Expression of endogenous HTT in mouse brain
In HD, the number of CAG repeats of the HTT gene determines the age of disease onset in a fully dominant fashion [
47]. When assessing the neuronal populations affected by pathogenic HTT protein aggregation, it is very likely that neurons with abundant HTT expression are more affected than neurons expressing low levels of HTT or no HTT at all. Thus, the defined cellular HTT expression levels might be decisive for the neuron type-specific pathology. From that point of view it is surprising that no detailed analyses on the neuron type-specific endogenous HTT expression patterns in brains of frequently used experimental animal species exist. There is only an early report showing HTT immunoreactivity in the cytoplasm of neurons in rat neocortex and in human cerebellar Purkinje and neocortical neurons as well as neurons of caudate nucleus [
17].
Therefore, we here comprehensively analyzed the brain region- and neuron type-specific expression of endogenous HTT in brains of three mouse lines, Wistar rat and hamster using a monoclonal rabbit antibody (EP867Y) that binds to the HTT mid-region. The specificity of this antibody is validated by Western blot analysis of HTT knock-out cells analyzed along with wild type cells (Product information sheet:
https://www.abcam.com/huntingtin-antibody-ep867y-ab45169.html). Furthermore, in the transgenic BACHD HTT mouse model we demonstrate much stronger EP867Y labelling than in wild type mouse brain tissue (Additional file
1: Figure S1b). In addition, the specificity of the EP867Y antibody was demonstrated in a number of recent publications [
32,
56,
75].
Interestingly, in the striatum of all mouse lines tested, as well as in rat and hamster, strong HTT immunoreactivity stood out in cholinergic interneurons that are spared in HD [
20]. In contrast, in DARPP-32-positive striatal medium-spiny neurons, which are known to be affected in HD [
29], significant HTT-labelling with the high threshold selected here was missing. However, also in human brain large striatal neurons that are spared in HD express HTT [
64] and deficits in cholinergic markers such as acetylcholinesterase and vesicular acetylcholine transporter without accompanying cell loss have been reported in HD affected subjects [
1,
73].
We detected ubiquitous low level HTT immunoreactivity throughout the brain but a highly selective enrichment of HTT in cholinergic neurons of the basal forebrain and amygdala, and in particular in cholinergic neurons of cranial nerve nuclei. In addition, prominent HTT expression was revealed in Ucn-1-positive EWN neurons and represented the only non-cholinergic neuronal population with abundant HTT expression identified in the present study. These observations indicate that not only neurons in striatum, but also defined neuronal populations in midbrain and brain stem may be affected by the pathogenic mechanisms resulting from polyQ extension of the HTT protein in HD. Indeed, in HD patients HTT protein inclusions were demonstrated in cranial nerve nuclei III (oculomotor nerve), IV (trochlear nerve), V (trigeminal nerve), VI (abducens nerve), VII (facial nerve), VIII (vestibulocochlear nerve), IX (glossopharyngeal nerve), X (vagal nerve) and XII (hypoglossal nerve) [
63]. This is in good accordance with our observation of prominent HTT expression in mouse, rat and hamster brain and would explain some clinical symptoms, (i.e. oculomotor and vestibular deficits) that cannot solely be attributed to degeneration of striatal neurons. Interestingly, HTT-rich vestibular projection neurons predominantly terminate on HTT-negative medium spiny neurons in dorsolateral striatum [
78], where significant neurodegeneration in HD occurs. Thus, in addition to local HTT expression, an indirect pathogenic HTT effect in striatum via innervating fibers should be considered based on the evidence presented here.
We would like to note that this pattern of HTT expression was not recapitulated by other HTT antibodies such as 2B7 and 2166. One possible explanation for this is HTT cleavage by different proteases to generate numerous fragments [
44,
57] and the adoption of different conformational states of the HTT protein [
65,
69]. The epitope of the EP867Y antibody is within amino acids 550 and 650 of HTT and corresponds to residues specific to the apopain (caspase-3) cleavage site [
27,
87]. Thus, this antibody appears to be unique and will allow monitoring distinct HTT expression related to cholinergic nuclei.
From the pattern of endogenous HTT expression in mouse, rat and hamster brain, one would predict that under conditions of pathogenic HTT expression the corresponding neuronal functions such as the reward system (VP, CPu, Tu, PPT, Amy), memory processing and decision making (Amy), arousal and attention (PPT), generation of REM sleep (LDT), pupil constriction and eye movement (EWN, oculomotor nerve nucleus, nucleus abducens), parasympathetic vagal functions (dorsal motor nucleus vagus) and general sensory and motor functions (facial nucleus, Pr5, Mo5, PPT, MVePC, ambiguus nucleus, hypoglossal nucleus) are affected. Indeed, the clinical features of HD such as progressive motor dysfunction, cognitive decline and psychiatric disturbance [
86] could be – at least partially – related to dysfunction or death of these neuronal populations. Typical motor dysfunctions in HD include chorea, dystonia, progressive bradykinesia, rigidity and incoordination. In addition, many patients have substantial cognitive or behavioral disturbances before onset of diagnostic motor signs [
54]. Interestingly, deep brain stimulation of the PPT might ameliorate gait and postural difficulties in PD subjects [
3,
79]. In addition, a significant cholinergic innervation of the striatum and nucleus accumbens arises from brainstem LDT and PPT nuclei [
14]. Thus, when affected during HD pathogenesis, dysfunction and/or degeneration of these neurons may contribute to striatal cholinergic deficits. Indeed, targeting the cholinergic system has been proposed as novel HD therapy [
18] and may compensate deficits in numerous cholinergic nerve nuclei with high HTT expression levels. This may also apply to the hypoglossal nerve nucleus which provides innervation of extrinsic and intrinsic muscles of the tongue and, therefore, accounts for the prominent dysphagia and disturbance of speech in HD patients [
48,
53,
62].
Subcellular localization
Neurons with highly abundant HTT immunoreactivity displayed cytoplasmic localization of HTT. This is consistent with biochemical and electron microscopic analyses which demonstrated the presence of HTT in neuronal cytoplasm and an association with vesicle membrane proteins [
17]. In the cytoplasm, full length HTT seems to be particularly involved in intracellular transport processes and vesicle trafficking, as it interacts with microtubules and clathrin-coated vesicles [
24,
36,
82]. In neurons with lower endogenous HTT expression levels, we also observed faint immunoreactivity in the nucleus. This is in line with reported co-immunoprecipitates with the carboxy-terminal binding protein, known to be a transcriptional co-repressor in the nucleus [
43]. An amino-terminal HTT fragment, however, has also been shown to interact with a variety of proteins important for nuclear function, like p53 [
77] and nuclear receptor co-repressor protein [
5].
Transgenic HD models
In several transgenic mouse models of HD, mutant HTT is randomly inserted within the genome and its expression is typically driven by neuron-specific promoters other than the endogenous HTT promoter. Most of these mice develop ubiquitinated nuclear HTT inclusions but do not exhibit the full neuropathological phenotype of human HD brains [
83]. In particular, brain regions affected by neuronal loss differ from that in HD and the effects on oligodendrocytes and astrocytes in the human brain are not recapitulated in transgenic mice [
83]. On the other hand, in
knock in mouse models of HD, transgene expression is driven by the endogenous mouse HTT promoter and should lead to pathology in brain regions identified by us as highly HTT expressing. Interestingly, the manifestation of HD-typical pathology including brain atrophy, striatal neuronal loss, severe motor phenotype, weight loss and premature death depends on the expression of N-terminal HTT fragments as present in the R6/2 model expressing exon 1 of the HTT protein under the control of human HTT promoter [
8,
49]. Unfortunately, the more caudal brain regions containing cholinergic cranial nerve nuclei identified by us as strongly immunoreactive for endogenous HTT were not included in the analyses. Thus, we strongly recommend the analyses of these structures and would predict the detection of pathological alterations like HTT protein aggregates and astrogliosis in these clearly defined nuclei.
Protein cross seeding
An important aspect of the present work was the analyses of a potential HTT protein cross seeding by Abeta aggregates in brains of transgenic mice with amyloid pathology. As the amyloid pathology in the Tg2576 mouse model used is restricted to neocortex and hippocampus, we focused on HTT aggregation in proximity to amyloid plaques in these brain regions. We observed an extracellular halo of HTT immunoreactivity around amyloid plaques labelled with the Abeta-specific antibody 4G8 or by the fibril intercalating dye ThS, respectively. Such protein cross seeding events are known to occur for proteins affected by a particular disease, such as Abeta and tau proteins in AD [
28], but also for proteins from different clinical entities such as Abeta and α-synuclein proteins affected in AD and PD [
23,
46,
52,
55,
80], respectively. We here demonstrate for the first time that also the endogenous HTT proteins can cross-seed with Abeta aggregates. This process has a clear age-related component and Abeta plaque-associated HTT immunoreactivity in hippocampus increases with plaque load. Since Abeta plaques also contain non-Abeta peptides that could be responsible for HTT recruitment, we directly tested the aggregation of HTT by Abeta (1-42) in ThT aggregation assays. We demonstrate that Abeta peptides increase the efficiency of the polymerization event but have no effect on the nucleation process. The nucleation of polyQ HTT was reported to underlie a different mechanism than nucleation of other amyloid peptides [
38]. PolyQ fibrils of HTT do not reveal a typical β-sheet structure but rather assemble into α-helical structures and oligomers which interfere with polyQ sequences and facilitate the formation of amyloid nuclei [
58]. A similar effect could occur in the presence of Abeta peptides generating conditions that favor β-sheet structure and thus result in higher aggregation rates [
74]. The biological significance and pathogenic potential of these co-aggregates is not yet clear but they might represent novel therapeutic targets. It has been recently shown in an AD mouse model that targeting one pathogenic protein by passive immunization also attenuates the deposition of other pathogenic protein aggregates, most likely by a mechanism involving activation of the complement system and microglia [
13].
Currently, no data are available on cross seeding of HTT to Abeta plaques in AD brains. However, there are several reports analyzing Abeta pathology in HD brains [
15,
40,
76]. Based on morphological amyloid and tau staging in autopsy cases of HD, a rare co-existence of HD and AD was reported, although initial neuropathological stages of AD were found to be present early in HD patients [
40]. On the other hand, in elderly HD subjects with dementia, a high proportion of subjects displayed tau and amyloid pathology, suggesting that co-occurrence of AD may contribute to cognitive decline in elderly HD patients [
15]. Moreover, a co-occurrence of mixed proteinopathies involving oligomerization and aggregation of Tau, α-synuclein and TDP-43 in late stage HD was demonstrated [
76], indicating common mechanisms of pathological protein aggregation in different neurodegenerative diseases.
HTT expression by astrocytes
In brain sections of Tg2576 mice used for the analyses of HTT protein cross seeding to amyloid plaques we surprisingly observed HTT immunoreactivity in reactive astrocytes in proximity to Abeta plaques. This robust astrocytic HTT immunoreactivity raises the question whether HTT produced by and released from neurons is taken up by astrocytes or whether it is expressed by these glial cells. In order to address this issue, primary neuronal and glial cell cultures from APP-transgenic Tg2576 mice were established and analyzed for Htt mRNA and protein expression. Both, primary neurons and astrocytes were shown to express Htt mRNA and protein. This is in agreement with immunohistochemical labellings of brain slices and supportive of de novo synthesis of HTT by astrocytes, rather than uptake of neuron-derived HTT. However, in contrast to the labelling of mouse brain sections, also astrocytes derived from primary cultures of wild type mice expressed HTT indicating that astrocytic HTT expression can generally be induced under conditions of astrocytic activation and independent of Abeta pathology. HTT immunoreactive astrocytes were also detected in the caudate and putamen of HD subjects [
72] and in a transgenic HTT mouse model [
71]. The intentional and specific mutant HTT expression by astrocytes was demonstrated to cause age-dependent neurological symptoms including body weight loss, motor function deficits, increased susceptibility to glutamate-induced seizures and early death [
6,
7]. Moreover, the mutant HTT expression by cultured primary astrocytes was shown to induce cell death of co-cultured wild type neurons [
71]. On the other hand, reduction of astrocytic mutant HTT expression slows disease progression in the BACHD conditional HD mouse model and improves motor and psychiatric-like phenotypes [
89]. Thus, there is solid evidence to conclude that astrocytic HTT expression contributes to neurodegeneration in HD.