N-methyl-D-aspartate (NMDA); neuronal intranuclear inclusion (NII). aggregates (11, 27). Postmortem studies of HD brains also show differential loss of projection neurons containing enkephalin, adenosine A2a, and dopamine D2 receptors compared with cells containing substance P, dynorphin, and dopamine D1 receptors (29–31). In juvenile HD, both types of striatal projection neurons are equally affected (32). In the cortex, neurons in the deeper layers (layers V and VI), which use the neurotransmitter glutamate, develop nuclear and neurite aggregates (27). Aggregates in HD were first observed in an electromicroscopy study of in vivo biopsies of HD brains (33). This observation was not pursued at the time, but it was remembered in 1996 when studies of the first HD transgenic mouse (28)(expressing the first exon of human huntingtin driven by the huntingtin promoter) were reported. These mice develop normally until around 5 weeks of age, when they begin to lose weight and to perform less well on the Rotorod test. Both brain weight and body weight diminish subsequently. The animals develop diabetes and tremors and become less active. They are finally moribund and die at around 13 weeks (34). Extensive early studies of the brains of these animals showed no clear neurochemical abnormalities like those seen in postmortem HD brains. Electromicroscopy studies, however, showed intranuclear inclusions. In the transgenic mice, it then became obvious from immunocytochemical studies that these NIIs were positive for the HD protein and for ubiquitin. Furthermore, virtually all neurons in the brains of these so-called R6/2 mice contained NIIs (28). These studies led scientists to revisit the examination of human brains in which NIIs were also found (11, 26). The frequency of NIIs in human HD brains was lower than in the transgenic mice, and the aggregates appeared as described above. The formation of aggregates was subsequently thought to be the sine qua non of HD pathogenesis. Two papers then appeared that suggested that the aggregates were an epiphenomenon, since cell death did not necessarily result from neuronal huntingtin aggregation, yet cell death did arise after the expression of mutant huntingtin (35, 36). It appeared that, to be toxic, the mutant protein had to get into the nucleus, since constructs with nuclear-export signals attenuated death resulting from exon 1 overexpression. Furthermore, inhibition of caspases rescued cells from death. Advocates of the hypothesis that aggregates cause cell death have been studying the phenomenon in various in vitro and cell culture assays (37, 38). Polyglutamine peptides are not soluble and need to be tied to other proteins to be studied in solution. Tight aggregates apparently form into polar zippers that are held together by hydrogen bonding (39). Some have also hypothesized that transglutaminases link glutamines to lysines covalently (40). No one has yet purified enough of the N-terminal fragment expressing the polyglutamine stretch to perform crystallization. The purified fulllength protein has not been isolated. Aggregation, however, can be studied by fusing a polyglutamine peptide with a GST protein with a sequence that can be broken with trypsin (41). Once the protein is in solution, the GST protein is cleaved off with trypsin, and then the aggregation process can be followed by filter assays. Aggregation is dependent on the length of the polyglutamine repeat. A transition seems to occur in the range of 39–40 repeats. Peptides with fewer than 39–40 glutamines aggregate less robustly than peptides with more than 40 repeats. Cell culture models also show length-dependent …