This year (2013) marks the 20th anniversary of identification of the causal genetic mutation for Huntington's disease (HD), a landmark discovery that heralded study of the biological underpinnings of this most common dominantly inherited neurodegenerative disorder. Among the variety of model organisms used to study HD pathogenesis, the mouse model is by far the most commonly used mammalian genetic model. Much of our current knowledge regarding mutant huntingtin (mHtt)-induced disease pathogenesis in mammalian models has been obtained by studying transgenic mouse models expressing mHtt N-terminal fragments, full-length murine or human mHtt. In this review, we focus on recent progress in using novel HD mouse models with targeted mHtt expression in specific brain cell types. These models help to address the role of distinct neuronal and non-neuronal cell types in eliciting cell-autonomous or non-cell-autonomous disease processes in HD. We also describe several mHtt transgenic mouse models with targeted mutations in Htt cis-domains to address specific pathogenic hypotheses, ranging from mHtt proteolysis to post-translational modifications. These novel mouse genetic studies, through direct manipulations of the causal HD gene, utilize a reductionist approach to systematically unravel the cellular and molecular pathways that are targeted by mHtt in disease pathogenesis, and to potentially identify novel targets for therapy.
Huntington's disease (HD) is an autosomal dominant neurodegenerative disorder characterized by progressive neurological symptoms, including involuntary movement (e.g. chorea, dystonia and gait abnormalities), cognitive deficits and psychiatric manifestations. Neuropathology reveals selective neurodegeneration and neuroinflammation occurring foremost in the striatum and to a lesser extent in the cortex . HD is caused by an expansion of CAG repeats translated into an elongated polyglutamine (polyQ) repeat near the N-terminus of the huntingtin (Htt) protein. While the polyQ length normally ranges from 6 to 34 repeats, individuals with 40 or more repeats invariably develop HD . The CAG repeat length correlates inversely with the age of motor symptom onset, a criterion for disease diagnosis . A critical but as yet unanswered question in HD is how the widely expressed mutant huntingtin (mHtt) protein causes age-dependent, selective neurodegeneration. Despite ubiquitous expression of mHtt in the brain and body, the striatal medium spiny neurons (MSNs) and deep-layer cortical projection neurons (CPNs) are the most vulnerable to degeneration in HD patients . A related question is whether degeneration of these specific neuronal types is due to purely intrinsic toxicity of mHtt in these neurons (cell-autonomous toxicity), or whether mHtt acting in other cell type(s) contributes to the demise of these vulnerable neurons (i.e. non-cell-autonomous toxicity). The third, and arguably the most pressing, question is what molecular targets, beyond mHtt itself (mRNA or protein), may be engaged to prevent or slow the disease process.
The onset of the plethora of behavioral and neuropathological phenotypes in HD patients takes decades, and the disease relentlessly progresses for 10–20 years from motor symptom onset to patient's death. To elucidate causal, mHtt-induced disease mechanisms during the long disease course from pre-symptomatic to symptomatic phases, it is insufficient to perform patient studies alone, and development of genetic animal models that recapitulate salient genetic, behavioral and neuropathological features of HD is required. If such models are available, a ‘two-step reductionist’ approach may be employed to unravel cellular logic and elucidate molecular mechanisms underlying selective pathogenesis in the disease model. The first step is to determine in which brain cell types expression of mHtt is necessary or sufficient to elicit critical aspects of the disease, either through cell-autonomous or non-cell-autonomous mechanisms. The second step is to further identify the precise molecular mechanisms within HD-relevant cell types, as defined by the first step, that modify the course of the disease. In this review, we summarize recent progress in mouse genetic studies on HD pathogenesis that fits this overall strategy. In particular, we focus on novel mouse genetic studies that systematically manipulate mHtt itself to answer questions related to the spatio-temporal requirement for mHtt expression in eliciting HD-like phenotypes in mouse models, and on novel mouse models that aim to address the impact of specific huntingtin cis-domains or post-translational modifications (PTMs) on disease pathogenesis.
Genetic mouse models of Huntington's disease: a brief update
Htt is a large protein consisting of 3144 amino acids. Beyond the first exon, which contains the polyQ stretch, Htt consists of dozens of HEAT repeats that are thought to mediate protein–protein interactions. Deletion of the huntingtin (Hdh) gene in mice leads to embryonic lethality [4-6], while conditional knockout of Hdh in post-mitotic neurons leads to progressive neurodegeneration in the cortex of aged mice . These data support an essential role for Htt in both embryonic development and adult neuronal survival. Although it remains plausible that partial loss of normal Htt function may contribute to HD pathogenesis , expanded polyQ mHtt fully substitutes for endogenous wild-type Htt during embryonic development, and hence mHtt gain-of-function toxicities are probably key to the disease process [9-11]. Consistent with this conclusion, it has been demonstrated through genome-wide gene expression analysis that polyQ-length-correlated and Hdh-null-altered genes act either in the same pathways or in inter-connected pathways. Close examination of energy and lipid metabolism categories revealed that mHtt-induced gene expression changes were distinct from, but related to, the effects of the lack of Htt, suggesting a gain-of-function mechanism through polyQ expansion .
There are a plethora of genetic mouse models in which mHtt is expressed (Table 1) , which together have been instrumental in increasing our understanding of many aspects of HD pathogenesis. By and large, there are two major types of HD mouse models differing in their genetic constructs to encode a fragment versus full-length Htt. The first are mHtt mouse models that express a toxic N-terminal fragment of mHtt. Indeed, the first in vivo evidence supporting the hypothesis that the polyQ expansion in an Htt fragment elicits neurotoxicity in mice was obtained through over-expression of a human mHtt exon 1 fragment with 144 Q (i.e. R6/2 and R6/1) . The R6 models were instrumental in discovery of in situ formation of aggregates by mHtt N-terminal fragments , which were subsequently confirmed in HD patient post-mortem brains . The subsequent development of additional mHtt fragment models, such as the N171-82Q model expressing the N-terminal 171 amino acids of mHtt with 82 Q , was key to demonstrating the reproducible and robust toxicities of mHtt N-terminal fragments. These fragment models often exhibit progressive and severe motor impairment, global brain and body weight loss, and prominent intra-nuclear and neuropil mHtt aggregation, features that are similar to those in patients. However, fragment models often have brain atrophy that is relatively global, and with only modest neuronal loss [18-20]. Certain features of the fragment models, such as the early age of onset of behavioral symptoms, rapid and often lethal disease progression, and presence of seizures in the R6/2 model, suggest that these models may capture certain features of juvenile-onset HD [13, 21].
Table 1. Summary of commonly used genetic mouse models of HD
As HD pathogenesis in patients is elicited by full-length mHtt over decades, two types of full-length mHtt models have been developed to study such slowly progressive pathogenic processes and are being increasingly used to study disease pathogenesis. One type of the full-length model is the knock-in (KI) models, in which expanded CAG repeats or human mutant HTT exon 1 are used to replace corresponding sequences in the endogenous murine Hdh locus . A series of mHtt-KI models with increasing polyQ length repeats are available, with Hdh-Q111 , CAG140 , Q150 [24-26] and zQ175 [27, 28] being the models most commonly used for HD molecular pathogenesis and therapeutic research . This allelic series of mHtt-KI models has the most precise genomic context, with the mice showing endogenous levels of full-length mHtt expression . However, it should be noted that KI mice express a hybrid of mostly murine Htt protein with human mHtt exon 1, under the control of murine promoters and genomic regulatory elements, and hence there are subtle differences at the levels of Hdh/HTT genomic DNA and protein sequences between KI mice and HD patients [13, 30]. The KI mice are valuable to study progressive mHtt accumulation and aggregation  and molecular changes in affected striatal and cortical neurons . Several mHtt-KI models, particularly those containing human HTT exon 1 and with repeats ranging from 140 to 200 Q, exhibit multiple behavioral deficits and late-onset brain atrophy that is mostly restricted to the same brain regions affected in patients, i.e. striatum and cortex [23, 32-35].
Although the majority of KI mice exhibit behavioral deficits and degenerative pathology at relatively late time points, a recently characterized mouse KI model, zQ175, has been shown to exhibit early and robust disease phenotypes [27, 28]. The zQ175 mice were obtained as a result of spontaneous expansion of the CAG repeat to approximately 175 from the previously generated CAG140 KI mice . Both homozygotes and heterozygotes were characterized, with the homozygotes developing motor and grip-strength deficits as early as 4–8 weeks, and progressive behavioral impairment, including rotarod and cognitive deficits, at 6 and 10 months of age. Heterozygous zQ175 mice showed locomotor deficits as early as 4.5 months of age, and striatal gene expression changes at 6 and 10 months . Importantly, both homozygotes and heterozygotes exhibit progressive whole-brain, cortical and striatal volume loss at 3 and 4 months of age, with evidence of striatal neuronal loss starting at 4.5 months of age in the homozygotes (approximately 15% reduction by stereological cell counting) . Additional characterization of the model also revealed progressive hyper-excitability in striatal MSNs in acute brain slices, and changes in magnetic resonance spectroscopy consistent with those seen in patients. As brain volume loss in homozygous but not heterozygous zQ175 mice occurs relatively early (e.g. 6 weeks of age), it remains possible that two copies of Hdh with 175 Q may elicit some developmental deficits . Overall, the gene dosage-dependent, relatively early and progressive behavioral, electrophysiological, pathological and molecular changes support use of this new KI model for HD pathogenic and pre-clinical studies.
The other type of HD mouse model is the mice that express full-length mHtt from the human genomic locus using either a yeast artificial chromosome (YAC) [36, 37] or a bacterial artificial chromosome (BAC) . Both YAC and BAC HD models introduce a large (> 200 kb) segment of the human HTT genomic locus into the mice, providing relatively intact human genomic regulatory elements and protein context within the span of the transgene. The YAC HD model lines, including YAC18, YAC46, YAC72 and YAC128, are named after the size of the polyQ repeat in the human HTT gene, while BACHD mice express mHtt harboring 97Q. As a group, the full-length human mutant HTT genomic transgenic mouse models demonstrate slowly progressive but relatively robust motor dysfunction (i.e. rotarod deficits), psychiatric-related behaviors and cognitive deficits, and selective atrophy in the striatum and cortex that spares the cerebellum [9, 36, 38]. The latter pattern of brain atrophy is reminiscent of that seen in HD patients. A major difference between YAC128 and BACHD mice is the DNA sequence encoding the polyQ sequence: YAC128 has a relatively pure CAG repeat, while BACHD has a mixed CAG/CAA repeat. The latter repeat is genetically stable in germ-line and somatic tissues, including the striatum, cortex and cerebellum, demonstrating that repeat instability, at least at a relatively long repeat range, is not necessary for full-length mHtt to elicit disease in this model . The BACHD mice have also been shown to exhibit relatively robust motor deficits and brain atrophy, as measured by power analyses of several phenotypic read-outs, while YAC128 mice generally show milder deficits [9, 34, 37]. These may be due in part to the relatively high levels of mHtt expression (i.e. an estimated 75% transgene expression in YAC128 versus 150–200% in BACHD compared to the endogenous murine Htt protein levels). However, even with higher mHtt expression, BACHD mice exhibit fewer mHtt aggregates than YAC128 mice do [9, 39]. The discrepancy between the level of aggregates and severity of behavioral deficits suggests that the aggregates detected by these antibodies (S830 and EM48) may not represent a toxic species that contributes to disease pathogenesis [39, 40]. A shared phenotype commonly observed in these human HTT genomic transgene mice is body weight gain, which is not observed in HD patients and is probably due to the Htt dosage effect, as YAC or BAC mice over-expressing wild-type Htt also show weight gain  (X Gu and XW Yang, unpublished data). An alternative explanation is that the hypothalamic toxicities elicited by mHtt lead to weight gain in these models . Several studies suggest that the behavioral deficits in HD genomic transgenic models cannot be simply attributed to body weight changes [9, 34]. Moreover, these mice exhibit HD-like brain and testicular atrophy [9, 43], demonstrating HD-like pathogenic processes in these models. Despite the overall similarities in behavioral deficits and selective brain atrophy, several other differences between BACHD and YAC128 models, including the timing and extent of nuclear mHtt accumulation and striatal gene expression changes , may be due to the levels of mHtt expression or the nature of repeats encoding the polyQ sequence. However, further investigation by our group (J.P.C. and X.W.Y., unpublished data) and others  found that 12-month-old BACHD mice showed similar reductions in expression of multiple striatal genes (e.g. Actn2, Ppp1r1b/Darpp-32, Ddit4l and Pcdh20), which are comparable to those altered in YAC128 striata . Thus, these data suggest molecular, behavioral and pathological phenotypic similarities between YAC128 and BACHD mice.
Conditional expression of mutant huntingtin to illuminate cellular targets in HD pathogenesis
In the brain, complex neural circuitry regulates normal functions of neurons in various regions. In HD patients, severe and progressive neurodegeneration is observed in two highly connected regions (the cortex and striatum), suggesting that cell–cell interaction among MSNs and CPNs may play an important role in HD pathogenesis. To address this question, we developed a conditional mouse model, called RosaHD, in which expression of an mHtt exon 1 fragment with 103 Q may be precisely turned on in specific cell types [46, 47]. To achieve conditional expression, mutant HTT exon 1 was targeted to the murine Rosa26 locus with two LoxP sites flanking a transcription termination sequence that is strategically placed before the HTT exon 1 sequence. Thus, expression of this toxic mHtt fragment is dependent on Cre-mediated excision of the stop sequence. By comparing RosaHD mice crossed with a line expressing predominantly striatal-selective Dlx5/6-Cre (which is also expressed in a population of cortical interneurons), a line expressing predominantly cortical glutamatergic neuron-selective Emx1-Cre, or a line expressing pan-neuronal Nestin-Cre, we were able to assess the relative contributions of cell-autonomous versus non-cell-autonomous toxicities of mHtt exon 1 in disease pathogenesis in the cortex and striatum. We found that mHtt exon 1 forms aggregates in a cell-autonomous manner in both the cortex and striatum. However, motor deficits and neurodegenerative pathology (e.g. dystrophic neurites, dark degenerating neurons, and reactive gliosis) were only observed in the cortex and striatum of RosaHD/Nestin-Cre mice at 1 year of age, and not in RosaHD/Emx1-Cre mice (selective expression of mHtt exon 1 in CPNs) and RosaHD/Dlx5/6-Cre mice (selective expression of mHtt exon 1 in striatal MSNs and cortical inter-neurons). These studies suggest, at least in the context of mHtt exon 1-induced disease, that purely cell-autonomous toxicities in either the cortical or striatal neurons alone are not sufficient to induce the full extent of the disease within the vulnerable neuronal populations, and full-scale neuropathogenesis in either the CPNs or MSNs requires non-cell-autonomous neuronal toxicities derived from mHtt expressed in other neuronal or glial cell types. The precise cellular origins of such toxicities were not defined in these initial studies, but they provide strong support to the notion that non-cell-autonomous toxicity may be relevant in HD pathogenesis.
The question of cell autonomy versus non-cell autonomy in HD pathogenesis is far from being fully understood. Our original studies of RosaHD/Emx1-Cre and RosaHD/Dlx5/6-Cre mice suggested that even cell-autonomous expression of mHtt exon 1 in CPNs or MSNs is able to yield modest toxicities, analyzed by sensitive methods such as electron microscopy or electrophysiology [46, 47]. Another study reported showed that transgenic mice (namely DE5) that express an N-terminal 171 amino acid fragment of mHtt with 98 Q selectively in MSNs (via the DARPP-32 promoter) show cell-autonomous mHtt aggregation, forebrain atrophy, age-dependent motor deficits and striatal gene expression changes [48, 49]. These studies support cell-autonomous toxicity of this mHtt fragment in MSNs. However, compared to another transgenic mouse model with pan-neuronal expression of the same mHtt fragment driven by the prion promoter, N171-82Q, the striatal pathology and overall disease phenotypes of N171-82Q mice are much more severe than those of DE5, despite the higher level of striatal mHtt fragment expression in DE5 mice . Overall, the studies of RosaHD model series and DE5 mice support the likelihood of contributions from both cell-autonomous and non-cell-autonomous toxicities in striatal and cortical pathogenesis in HD fragment models.
Glial cells play an essential role in maintaining brain homeostasis to support proper neuronal function. In HD, reactive astrogliosis and microgliosis are consistently found in the striatum and cortex of post-mortem HD brains [50, 51]. Furthermore, elevated pro-inflammatory cytokines such as interleukin-6 and tumor necrosis factor α are found in the blood and cerebrospinal fluid of HD patients , suggesting that peripheral and central inflammatory reactions may mark or actively contribute to ongoing disease processes . The activation of astrocytes and microglia may be a double-edged sword, possibly having protective and harmful roles in neurodegenerative diseases . An elegant example of glia or microglia contributing non-cell-autonomously to neuropathogenesis in the case of familial amyotrophic lateral sclerosis elicited by mutant superoxide dismutase 1 has been published [55-57]. In HD, as mHtt is broadly expressed in both neuronal and non-neuronal cells, and mHtt aggregates and glial dysfunction, such as reduced glutamate transporter expression, have been found in HD mouse models, it is conceivable that glial mHtt may elicit cell-autonomous or non-cell-autonomous disease processes in HD. To address such a question, Bradford et al. developed a transgenic mouse model with astrocyte-specific expression of a 208 amino acid N-terminal fragment of mHtt (N208) with 160 Q repeats . These mice develop age-dependent neurological phenotypes, including motor impairment, body weight loss and reduced survival, while mice expressing the N208 fragment with a short polyQ repeat (23 Q) are spared the disease. The mutant mice exhibit pathology such as mHtt aggregation and gliosis, but do not show neuronal death. These results suggest that mHtt in astrocytes contributes to neurological symptoms through dysfunction rather than degeneration of the neuronal systems. A second study by the same group of investigators further strengthens this view by showing that astrocyte-specific expression of an mHtt N208 fragment with 98 Q by itself does not show overt phenotypes, but may exacerbate disease caused by expression of prion promoter-driven N171-82Q in double transgenic mice . These studies provide compelling evidence that mHtt fragments in astrocytes contribute to HD pathogenesis.
One caveat of the published studies examining the cell type-specific roles of mHtt in HD is the exclusive use of mHtt fragment models, which may not have ideal construct validity when compared to the full-length mHtt mouse models . Therefore, studies using mouse models with conditional expression of full-length mHtt in specific neuronal or non-neuronal cell types in genetic mouse models of HD are required to further study the cellular targets of mHtt in HD and provide new insights into how full-length mHtt elicits pathological interactions between cortical and striatal neurons to contribute to disease pathogenesis.
Mouse genetic models to study mutant huntingtin proteolysis and post-translational modifications in HD pathogenesis
Another useful approach to assess HD pathogenesis using mouse genetic models is to develop models carrying specific mHtt cis-domain mutations that reside outside the polyQ domain. These mutations either mimic or block potential pathogenic events that directly affect mHtt itself. Such studies may provide rigorous evidence that a particular Htt domain or its PTM affect the plethora of pathogenic consequences resulting from expression of mHtt. HTT is a very large gene, covering a 170 kb genomic region in human and encoding a protein with > 3144 amino acids. Htt has been shown to be capable of interacting with several hundred mammalian brain proteins [61, 62]. Therefore, Htt cis-domains, their PTMs and/or interacting proteins may play crucial roles in the production, trafficking, function and clearance of Htt itself. A precise understanding of such mechanisms, and a thorough evaluation of their role in HD pathogenesis, may suggest a rational strategy towards Htt-targeted therapy for HD.
A rigorous line of research in this area is the study of mHtt proteolysis, which creates smaller and more toxic mHtt N-terminal fragments that are known to accumulate as nuclear inclusions and cytoplasmic/neuropil aggregates in the brains of HD patients and mouse models [15, 63, 64]. Full-length mHtt may be cleaved into a large number of N-terminal fragments, and a subset of the enzymes that generate N-terminal mHtt fragments of varying sizes have been determined . They include caspases 3 and 6 (Casp3 and Casp6, respectively) [66, 67], calpain , matrix metalloproteinase 10 , and an as undefined aspartyl protease . Interestingly, a recent study by Sathasivam et al. also showed a potential genetic mechanism to produce an mHtt exon 1 product independent of proteolysis . This study showed that aberrant splicing of mutant HTT exon 1, in a CAG repeat length-dependent manner, yield a short polyadenylated HTT mRNA transcript that is translated into an mHtt exon 1 protein product. The precise contribution of many of the distinct proteolytic mechanisms or the newly identified alternative splicing mechanism to the overall disease pathogenesis remains to be clarified. One of these mechanisms, cleavage of mHtt into an N-terminal 586 fragment of mHtt, has been investigated using mouse genetic models. Casp6 cleaves mHtt at residue 586 , with the subcellular compartment for such cleavage appearing to be the nucleus . The original mouse genetic experiment demonstrating the significance of Casp6 cleavage in HD pathogenesis, performed by Graham et al., demonstrated that YAC128 transgenic mice expressing a genomic copy of human full-length mHtt carrying a mutation that blocks Casp6 cleavage show prevention of disease pathogenesis compared to the baseline YAC-128Q model, with such an effect not being seen in YAC mice with mHtt mutations that blocked Casp3 cleavage . Activation of Casp6 is an early pathogenic event in HD mice, and its level of activation is directly correlated with CAG repeat length and inversely correlated with the age of onset in HD patients . Recent studies have shown that transgenic expression of mHtt caspase fragments in mice elicits disease that appears to be more severe than that for full-length mHtt models but less severe than that for models expressing smaller mHtt fragments, with these mice accumulating predominantly cytoplasmic mHtt aggregates [75, 76]. Thus, these transgenic experiments confirm the toxicity of the Casp6 fragment of mHtt, but suggest that this cleavage product may be an intermediate that may be further processed into smaller and probably more toxic mHtt fragments. These studies provide proof-of-concept that specific mHtt cleavage events, at least through cis-mutations, prevent the onset of both neuronal dysfunction and neurodegeneration in HD mice expressing full-length mHtt.
Recent advances in studying the role of Casp6 in HD pathogenesis have moved towards genetic validation of Casp6 itself as a potential target for HD therapy, primarily by crossing various HD mouse models with Casp6-deficient mice [65, 77]. However, so far, these studies do not indicate that Casp6 alone is responsible for mHtt cleavage at residue 586, as crossing Casp6 null alleles into either HdhQ150 or BACHD mice did not show reduced levels of cleaved mHtt fragments at this site, and extracts from Casp6 null mouse brains are still able to generate the mHtt 586 amino acid fragment from full-length mHtt [77, 78]. Unexpectedly, BACHD/Casp6−/− mice showed a reduced overall level of mutant and wild-type Htt, and modest improvement of behavioral deficits but no effect on neurodegeneration . The modest benefit of Casp6 reduction is correlated with activation of mHtt clearance, probably through autophagy and the ubiquitin proteasome systems. Taken together, the data so far suggest that Casp6 cleavage site mutations around residue 586 may block the disease via inhibition of mHtt cleavage by several potential proteases, including Casp6. Therefore, further identification of the set of proteases that mediate this critical cleavage of mHtt is crucial to move forward on this path towards developing therapeutics. Moreover, it is equally important to evaluate, using mouse genetic means, the pathogenic significance of alternative mechanisms in generation of mHtt N-terminal fragments, such as matrix metalloproteinase 10  and aberrant splicing of mHtt exon 1 .
Another exciting avenue for investigation of the pathogenesis of HD and related polyQ disorders is use of novel mouse genetic models to assess the roles of PTMs on the disease-causing protein [30, 79, 80]. An elegant example of a PTM as an important mechanism in HD pathogenesis is acetylation of lysine at position 444, which is a signal to traffic mutant and wild-type Htt to the autophagosome for selective clearance . Importantly, rendering mHtt incapable of being acetylated (via a K444R mutation) results in dramatic accumulation of protein and enhanced neurodegeneration in cultured neurons and mouse brain, suggesting that boosting Htt-selective clearance mechanisms may be beneficial in HD.
Another domain of Htt that is being intensively studied in the context of PTMs in HD pathogenesis are the 17 amino acids of Htt (N17 domain) immediately preceding the polyQ domain . The N17 domain of Htt is highly conserved amongst all vertebrate Htt paralogs, but is not present in invertebrates . It forms an amphipathic α-helical structure [84, 85] that accelerates polyglutamine length-dependent aggregation in vitro [86, 87]. Several elegant studies have shown that an important cellular function of the N17 domain is to target Htt and its small N-terminal fragments to the cytosol, probably through distinct mechanisms that mediate association with cytoplasmic membranous structures (e.g. ER or mitochondria) [84, 88] and by facilitation of Crm1-dependent nuclear export [89, 90]. The precise molecular cascades underlying N17 domain-mediated nuclear–cytoplasmic shuttling of Htt, including its N-terminal fragments, and its relevance to HD, are currently under intensive investigation. Candidate N17 domain-interacting proteins such as 14-3-3 [91, 92], Tpr  and Tcp1  have already been implicated in modulating mHtt subcellular localization and/or aggregation. An impressive aspect of N17 domain biology, despite its small size, is the demonstration of more than 10 possible PTMs, including ubiquitination, SUMOylation, acetylation, phosphorylation and oxidation [84, 94-97]. The pathogenic significance of some of these PTMs has been explored in model organisms. For example, the ubiquitination and SUMOylation of the lysine residues K6, K9 and K15 appear to exert opposing roles in mHtt exon 1 toxicity in a Drosophila model . Our laboratory was the first to explore the pathogenic significance of serine 13 and 16 (S13 and S16) phosphorylation in the context of full-length mHtt-induced pathogenesis in HD mice . We generated BAC transgenic mice expressing full-length mHtt with 97 Q (the same as BACHD) with either phospho-mimetic (S13D and S16D; ‘SD’) or phospho-resistant mutations (S13A and S16A; ‘SA’). We demonstrated that both forms of full-length mHtt are able to rescue murine Hdh knockout lethality. SA mice reproduce all the key disease features originally observed in the BACHD model, including age-dependent motor deficits (rotarod and open field exploration), two psychiatric-related behaviors (depression-like and anxiety-like deficits) and selective forebrain atrophy, while two independent SD transgenic mouse lines did not exhibit these disease phenotypes at all [9, 98] (X Gu and XW Yang, unpublished data). The neuroprotective effects of the SD mutations or S13 and S16 phosphorylation have also been demonstrated in cell and mouse brain slice models [94, 97]. A surprising finding from our mouse study is that SD but not SA mice no longer exhibit mHtt aggregation at 12 months of age . This finding was supported by in vitro studies using mHtt exon 1 peptides with SD mutations or with phosphorylated S13 and S16 residues, which were both shown to block oligomer formation by mHtt fragments, leading to retardation of mHtt aggregation and inhibition of amyloid fibril formation [87, 98, 99]. Another explanation for reduced aggregation by S13 and S16 phosphorylation is the finding that these PTMs promote mHtt clearance via the ubiquitin proteasome and autophagy pathways . Together, these studies suggest that S13 and S16 phosphorylation may act as a molecular switch to critically modulate mHtt trafficking or clearance, and reduce mHtt toxicity in cell and animal models of HD. The precise kinases mediating N17 domain phosphorylation are being actively investigated, with candidate kinases including IKK  and CK2 . A long-term goal of this line of investigation, in addition to elucidating precise biological functions and regulatory pathways for the Htt N17 domain, is to develop new HD therapies. Encouraging developments include the discovery of small molecules that appear to boost S13 and S16 phosphorylation in vitro  and a small ganglioside compound (GM1) that exerts neuroprotection in YAC128 mice and concomitant enhancement of S13 and S16 phosphorylation on mHtt . The precise causal relationship between enhancement of N17 domain phosphorylation by these small molecules and disease suppression requires further investigation, and may potentially be demonstrated by differential effects of such compounds on HD mice expressing mHtt with a wild-type N17 domain or an N17 domain with phospho-resistant mutations (e.g. BACHD versus SA mice).
Since the discovery of HTT as the causal gene in HD, there has been a tremendous effort to determine the pathological role of mHtt at molecular, cellular and organism levels to obtain insights that may inform novel therapeutic strategies. At present, our sophisticated understanding of Htt biology and the advancement of tools to study in vivo disease proteins have led to development of a series of novel mouse models that allow molecular and spatial manipulations of mHtt itself. This new generation of mHtt-expressing models has allowed systematic and rigorous testing of pathogenic hypotheses related to age-dependent, progressive and selective neuronal pathogenesis in HD. The data so far support the use of Cre/LoxP conditional mouse models and cell type-specific promoter-driven mHtt models to address the cell-autonomous versus non-cell-autonomous toxicities of mHtt (Fig. 1A). Moreover, mouse models with mHtt cis-domain mutations, particularly those addressing specific pathogenic contributions of distinct PTMs, have also proven to be valuable in validating disease-modifying mechanisms (Fig. 1B). Together, by using the two-step reductionist approach, the cellular and molecular targets of mHtt in HD pathogenesis are starting to be determined, with such knowledge likely to offer new insights in HD therapeutic development.
The authors are grateful for support of X.W. Yang's laboratory by the US National Institute of Neurological Disorders and Stroke/National Institutes of Health (grants numbers R01NS049501 and R01NS074312), the CHDI Foundation, the Hereditary Disease Foundation, the David Weil Fund to the Semel Institute at University of California, Los Angeles, and a Neuroscience of Brain Disorders Award from the McKnight Endowment Fund for Neuroscience.