Aggregation of Huntingtin in Neuronal Intranuclear Inclusions and Dystrophic Neurites in Brain

See allHide authors and affiliations

Science  26 Sep 1997:
Vol. 277, Issue 5334, pp. 1990-1993
DOI: 10.1126/science.277.5334.1990


The cause of neurodegeneration in Huntington's disease (HD) is unknown. Patients with HD have an expanded NH2-terminal polyglutamine region in huntingtin. An NH2-terminal fragment of mutant huntingtin was localized to neuronal intranuclear inclusions (NIIs) and dystrophic neurites (DNs) in the HD cortex and striatum, which are affected in HD, and polyglutamine length influenced the extent of huntingtin accumulation in these structures. Ubiquitin was also found in NIIs and DNs, which suggests that abnormal huntingtin is targeted for proteolysis but is resistant to removal. The aggregation of mutant huntingtin may be part of the pathogenic mechanism in HD.

The pathology of HD is marked by a preferential loss of neurons in the striatum and cortex (1). The genetic mutation is an unstable and expanded CAG repeat in the gene that encodes huntingtin (2). Larger polyglutamine expansions in huntingtin are associated with earlier onset and increased severity of the disease (3). Because mutant huntingtin is expressed throughout the brain in HD (4), its involvement in selective cell death in the striatum and cortex is unclear.

Two pathogenic processes have been suggested as the basis for neurodegeneration in HD. One process involves interaction of mutant huntingtin with other proteins to produce a change of function. Alternatively, mutant huntingtin might homodimerize (5) or heterodimerize (6) to build large, poorly soluble protein aggregates. Proteins that interact more avidly with NH2-terminal products of mutant huntingtin than with wild-type have been identified but are found throughout the brain with no preferential distribution in those regions affected in HD (7). Analysis of the HD brain (8) with an antiserum that recognizes an internal region of huntingtin in wild-type and mutant proteins showed that the subcellular distribution of huntingtin in the cytoplasm of neurons was abnormal, but the contribution of mutant huntingtin to these changes was unclear. In a recent study of HD transgenic mice expressing an NH2-terminal mutant huntingtin fragment with 115 to 156 glutamine repeats, we found that intraneuronal nuclear inclusions reactive to NH2-terminal antiserum to huntingtin developed in the brain (9).

We therefore tested the hypothesis that abnormal aggregates of the NH2-terminal region of the HD protein accumulate selectively in neurons that degenerate in HD. We analyzed immunohistochemistry in postmortem brain tissue from controls (n = 5) and from HD patients with juvenile (n = 3) and adult onset (n = 6) HD (10). We used an antiserum to huntingtin (Ab 1) raised against an NH2-terminal epitope of huntingtin amino acids 1 to 17, which are proximal to its polyglutamine region, and compared these results with those obtained with an antiserum directed to an internal site at amino acids 585 to 725 in huntingtin (Ab 585) (11). Ab 1 has been characterized in biochemical and immunohistochemical studies of human and rodent brains (4,11) and in immunoblots it detects wild-type and mutant huntingtins in HD brain (4). In neurons of the HD cortex, Ab 1 produced intense labeling for huntingtin localized to neuronal intranuclear inclusions (hNIIs; Fig. 1A). hNIIs were positioned variably throughout the nucleus, adjacent to (Fig. 1B) or distant from the nucleolus (Fig. 1C). They were significantly larger (P < 0.0001; n = 65; mean = 7.1 ± 3.0) than the nucleolus (mean = 4.0 ± 1.6) in mean cross-sectional area. Compared with the nucleolus, which filled 0.8 to 18% of the cross-sectional area of the nucleus, hNIIs in about 30% of neurons covered 20 to 45% of nuclear cross-sectional area (Fig.2B). Analysis of the ratios of the major and minor axes of hNIIs (n = 245) revealed that about 55% were spherical, 30% were ovoid, and 15% were elliptical (12). One hNII per cell was most common but two or three per neuron were also seen in 5 to 7% of labeled neurons (Fig. 1C). Neurons with hNIIs were detected in all cortical layers and were more frequent in juvenile patients (38 to 52% of total neurons) than in adult patients (3 to 6% of total neurons) (Fig. 2A). They were not found in the cortex of adult patient A4, who was positive for the HD allele but presymptomatic at the time of death.

Figure 1

Huntingtin immunoreactivity in hNIIs and hDNs in HD brain. (A) Cortex of juvenile patient J13 shows numerous hNIIs prominently stained. (B and C) Cortical pyramidal neurons in juvenile patient J12 shown with Nomarski optics contain one (B) and two (C) hNIIs. The nucleolus in each cell is unlabeled. (D) Striatal neurons with hNIIs in juvenile patient J11. (E and F) hDNs in the cortex of adult HD patient A12 (E) and presymptomatic adult patient A4 (F), who had the HD gene. (G) Cortical neurons stained with Ab 585 show staining in cytoplasm but not in NIIs. A, bar = 50 μm; B–G, bars = 10 μm.

Figure 2

Analysis of hNIIs and hDNs in HD cortex. (A) Frequencies of hNIIs (□) and hDNs (▪) differ in juveniles and adults. Two other adult HD patients had results qualitatively similar to those of the adult patients presented here. (B) Percent of nuclear cross-sectional area occupied by hNII (□) and nucleolus (▪) is compared in HD cortical neurons. (C) Double-label immunofluorescence shows hDN (arrows) positioned within a neurofilament labeled axonal process.

hNIIs were also seen in medium-sized neurons in the striatum (Fig. 1D) but were not present in neurons of the HD globus pallidus or cerebellum. hNIIs were absent in the cortex, striatum, and other areas in brains of controls.

We found intense staining in extracellular structures that had a morphology consistent with dystrophic neurites (hDNs) (Fig. 1E). hDNs were present predominantly in cortical layers 5 and 6, where they were distributed unevenly in patches of neuropil and sometimes aligned in linear arrays reminiscent of processes. They were spherical or slightly ovoid and occasionally had thin extensions. Double labeling for huntingtin and neurofilament protein (10) showed that hDNs were contained within or continuous with neurofilament labeled axonal processes (Fig. 2C). hDNs had a mean length of 5.0 ± 1.7 μm (n = 256) and the largest were 10 to 12 μm. They were more prevalent in the cortex of patients with adult onset than in juvenile-onset patients (Fig. 2A). Some hDNs were detected in layer 6 cortex of the presymptomatic adult patient A4 (Fig. 1F). hDNs were seen in the HD striatum of adult and juvenile patients but they were absent from control brains.

Immunohistochemical analysis with Ab 585 showed labeling of the cytoplasm of neurons in control and HD brains (Fig. 1G) (8) but no staining of NIIs and DNs in neurons of the HD cortex (Fig. 1G), striatum, globus pallidus, or cerebellum. Altogether, the results suggested that the hNIIs and hDNs recognized by NH2-terminal antibody Ab 1 contained a cleaved fragment of mutant huntingtin not seen with Ab 585 (13). To further explore this idea, we examined nuclear extracts from the cortex of controls and juvenile HD patients by Western blot analysis (14). As expected in the controls, full-length huntingtin, which migrates at about 350 kD, was present in total protein extracts but not the nuclear fractions, consistent with the absence of nuclear localization of full-length huntingtin. A prominent band migrating at about 40 kD in total protein homogenates and in soluble nuclear extracts was detected in the HD cortex but not in the control cortex (Fig. 3). Together, our immunoblot and immunohistochemical data suggest that an NH2-terminal fragment of mutant huntingtin translocates to the nucleus and contributes to the formation of NIIs (15).

Figure 3

Western blot of huntingtin in control and HD cortex analyzed with NH2-terminal Ab 1. Full-length wild-type huntingtin in controls (C1, C8, C18, and C19) and wild-type and mutant huntingtins in juvenile HD patients (J6, J11, J12, and J13) migrate at about 350 kD (small arrow) in total protein homogenates (T). A fragment of about 40 kD (large arrow) is present in total protein homogenates (T) and soluble nuclear extracts (N) of HD patients but not in controls. Immunoreactive bands < 40 kD in the HD brain may be degraded products of the 40-kD fragment, other NH2-terminal fragments of huntingtin with different sites of cleavage, or a fragment of wild-type huntingtin. The nuclear fractions of patients J11 and J12 contain a small amount of full-sized mutant huntingtin, which suggests that uncleaved mutant huntingtin may also translocate to the nucleus. Isolation of nuclear proteins separate from cytoplasmic proteins is shown by the absence of α-tubulin in soluble nuclear extracts. Molecular mass markers are on the left.

Recent observations have shown that huntingtin can be cleaved in its NH2-terminal region by apopain, a cysteine protease involved in ubiquitin-dependent proteolysis, and that the rate of cleavage increases with the length of the polyglutamine tract of huntingtin (16). Because the NH2-terminus of mutant huntingtin is a substrate for apopain (16) and is ubiquitinated in lymphocytes (17) and because DNs containing ubiquitin have been observed in the HD cortex (18), we speculated that NIIs and DNs in HD cortical tissue would be detected with ubiquitin antiserum. We found NIIs (Fig. 4, A and B) and DNs (Fig. 4, C and D) with ubiquitin immunoreactivity in the HD cortex. Double-labeling for ubiquitin and huntingtin in the same section showed that the proteins were colocalized in NIIs and DNs (Fig. 4E). The frequency of ubiquitin-positive NIIs and DNs was directly proportional to the frequency of hNIIs and hDNs in adjacent brain sections from the same HD patients (Fig. 4F). However, there were usually fewer NIIs and DNs labeled with ubiquitin than with huntingtin (19). These results demonstrate that the mutant huntingtin aggregates in NIIs and DNs are ubiquitinated. Consistent with this finding are observations in HD transgenic mice that show the nascent nuclear inclusions contain huntingtin and ubiquitin immunoreactivity (9).

Figure 4

Ubiquitin and huntingtin immunoreactivity in NIIs and DNs in the HD cortex. (A andB) Ubiquitin-labeled pyramidal cells show one large and two small NIIs, respectively. (C andD) Large ubiquitin-positive DNs are shown, including one with a characteristic tail-like process. Bar = 10 μm. (E) Confocal immunofluorescence microscopy shows colocalization of huntingtin and ubiquitin in a NII (arrows, top) and in a DN (bottom). (F) Comparison of the frequency of NIIs and DNs, respectively, with huntingtin (□) and ubiquitin (▪) based on analysis of adjacent stained sections. Number of ubiquitin-labeled NIIs and DNs is directly proportional to but less frequent than the number with huntingtin in most HD patients.

Electron microscopic study showed that hNIIs were highly heterogeneous in composition and contained a mixture of granules, straight and tortuous filaments, and masses of parallel and randomly oriented fibrils (Fig. 5, A, C, and D). There was no membrane separating the hNII from the surrounding nucleoplasm. hDNs identified at the ultrastructural level contained labeled granules and filaments. A rim of cytoplasm surrounded the aggregate and contained an accumulation of organelles, especially mitochondria (Fig.5B). A granulofilamentous consistency has also been noted in nuclear inclusions identified in biopsy tissue from the HD cortex and striatum (20) and in cortical neurons of the HD transgenic mouse (9) as well as for ubiquitin-positive DNs of the HD cortex (18). Thus, based on ultrastructure the same mechanism may be involved in the accumulation of mutant huntingtin in NIIs and DNs (21).

Figure 5

Electron microscopy of hNIIs and hDNs in the HD cortex with immunoperoxidase labeling. (A) hNII in a cortical neuron appears as a dense aggregate with no limiting membrane separating it from the nucleoplasm. (B) hDN contains an aggregate of immunoreactive granules and filaments, which is surrounded by a rim of cytoplasm where mitochondria are accumulated. (C) Higher magnification of NII in (A) shows the presence of labeled granules and filaments within the inclusion. (D) Serial section of hNII in (A) and (C) shows fibrils organized in random and parallel arrays. Bars = 1.0 μm.

The presence of hNIIs in symptomatic HD patients and their absence in a presymptomatic adult favors the idea that hNIIs are closely linked to the onset of the disease. In accordance with the patient data, transgenic mice develop nuclear inclusions in the cortex and striatum (and in some other regions) just before the appearance of a neurological HD-like phenotype (9). The prevalence of hDNs in deep layers of cortex correlates with greater neurodegeneration in these layers (22), and their appearance in a presymptomatic adult suggests that they precede clinical onset. We found hDNs associated with neurofilament-positive axonal fibers, which agrees with evidence that DNs are distended axon terminals (23). The marked difference in occurrence of NIIs and DNs in juvenile and adult HD patients suggests that CAG repeat number influences development of these neuropathological features, which might account for the distinct clinical phenotypes of these two groups of patients (24). The shared features of other neurodegenerative diseases with CAG expansions and HD (25) suggest that the formation of nuclear inclusions and DNs may be a common pathogenic pathway.

Because brain regions affected in HD contained hNIIs and hDNs, the formation of these structures is directly implicated in HD pathogenesis. The irreversible aggregation of mutant huntingtin in one of the ways recently proposed (5, 6) would prevent its removal from cells. Neuronal dysfunction could arise because the aggregates physically interfere with the normal activities of the neuron or bind to and render inactive other polyglutamine-enriched proteins such as transcription factors in RNA synthesis or other huntingtin interacting proteins important for cell survival (7,26). The presence of ubiquitin in NIIs and DNs suggests that both structures are targets for ubiquitin-dependent proteolysis (27), although the less frequent occurrence of ubiquitin than of mutant huntingtin suggests that ubiquitin-dependent proteolysis is incomplete (28). Therapeutic approaches that inhibit aggregation of mutant huntingtin or increase the efficiency of its ubiquitin-dependent proteolysis may be helpful in the treatment of HD.


Stay Connected to Science

Navigate This Article