To address the role of pathological cell-cell interactions in striatal pathogenesis in HD mousemodels in vivo
, we developed a striatum-specific mouse model of HD which expresses a neuropathological fragment of mhtt (mhtt-exon1) selectively in striatal MSNs and in a subset of cortical interneurons but not in the CPNs or other neurons in the brain. This model reveals that striatal-specific expression of mhtt is sufficient to elicit cell-autonomous mhtt nuclear accumulation and aggregation and NMDA receptor dysfunction. However, cell-autonomous toxicity of mhtt is insufficient to elicit progressive locomotor deficits and significant striatal pathology including gliosis and dark neuron degeneration, all of which are observed in the pan-neuronal models of HD [27
Responses to NMDA and their blockade by Mg2+
were altered in mutant compared to WT mice in the pan-neuronal and striatal model, but not in the cortical model. This indicates that the presence of mhtt in CPNs alone does not lead to changes in NMDA receptor-mediated currents in the striatum. The changes in NMDA receptor function in the pan-neuronal and striatal models were different, with smaller NMDA currents in mutant cells from the pan-neuronal model, and larger NMDA currents in mutant cells from the striatal model. However, in both models Mg2+
blockade of NMDA currents was reduced, similar to effects in R6/2 mice of 15–40 days [31
]. Previous studies on dissociated neurons in R6/2 mice have shown both increases and decreases of glutamate receptor-mediated currents depending on the age and the progression of the phenotype. In young R6/2 mice (15–40 days), NMDA and AMPA currents were increased, while in older mice (80 days), they were decreased [31
]. Similarly, biphasic changes for NMDA and AMPA currents were observed in CPNs in young and older R6/2 mice [33
At the present time it is difficult to determine the mechanisms underlying this complex set of electrophysiological changes in different models at the same age. The presence of mhtt in striatal neurons appears to be necessary to alter Mg2+
sensitivity. This effect might be related to alterations in NMDA receptor subunits [34
]. The decrease in magnitude of the NMDA current in the pan-neuronal model might be related to changes in cortical inputs which have been shown to occur in this and other transgenic mouse models of HD [27
]. In contrast, the increase in current magnitude in the striatal model might reflect the possibility that alterations in the cortex no longer contribute to striatal anomalies.
The core observations of the present study are consistent with our prior analyses of the cortical model of HD. In both cases, we found that mhtt aggregation is cell-autonomous, but cell-autonomous toxicity of mhtt in either CPNs or MSNs is not sufficient to elicit robust regionally-specific neuropathology and behavioral deficits as compared to the pan-neuronal model. Thus, pathological cell-cell interactions appear to contribute to neuronal toxicity in both vulnerable neuronal populations in HD. Although cell-type specific expression of mhtt-exon 1 is insufficient to elicit robust pathology, our analyses did provide evidence of cell-autonomous neuronal dysfunction. In the cortical model, we showed expression of mhtt-exon1 in CPNs resulted in mild degenerative changes (i.e.
dark degenerating vacuoles in the CPNs under EM; see ref. [27
]); and in the striatal model, electrophysiological recordings revealed cell-autonomous NMDA receptor-mediated response changes in dissociated MSNs. Moreover, the regionally-specific expression of mhtt-exon 1 appears to be insufficient to elicit robust non-cell-autonomous toxicity in other brain regions. For example, selective expression of mhtt-exon1 in CPNs is not sufficient to cause a significant striatal phenotype in the cortical model. Thus, pathological cell-cell interactions are necessary but not sufficient for cortical and striatal pathogenesis in HD. Based on these observations, we propose a ''two-hit'' hypothesis in which mhtt can elicit both cell-autonomous toxicity and pathological cell-cell interactions which are necessary for selective neuronal toxicity in HD (Fig. )
Figure 6 Schematics illustrating the two-hit hypothesis of neuropathogenesis in conditional models of HD. Our current analyses of the striatal model and our previous study of the cortical and pan-neuronal model of HD have shown that behavioral deficits and robust (more ...)
Our analyses of the regional models of HD are consistent with prior evidence supporting the possible role of cell-cell interactions in HD striatal pathogenesis. First, excitotoxicity is thought to contribute to striatal pathogenesis in HD. The mechanisms underlying such toxicity are not completely clear, but may include both cell-autonomous mechanisms such as alteration in NMDA receptor function in the MSNs [17
] or non-cell-autonomous mechanisms, such as enhanced glutamate release from CPNs [40
] or reduced glutamate clearance [44
]. Our in vitro
electrophysiological analysis clearly confirms prior results with R6/2 mice [23
], and suggests that altered Mg2+
blockade of NMDA receptors in striatal MSNs may contribute to striatal excitotoxicity. Another mechanism supporting cell-cell interactions in HD is the role of cortical BDNF in striatal pathogenesis suggesting that the transcription and axonal transport of BDNF is diminished in HD models [25
]. Since cortex-specific deletion of BDNF is known to elicit progressive striatal degeneration in mice [26
], BNDF reduction will contribute to striatal pathogenesis in HD. Finally, non-neuronal cells, particularly astrocytes, are also affected in HD and HD mice. Studies using primary cell culture demonstrate that astrocytes derived from R6/2 mice have reduced capacity to transport glutamate, and thus cannot protect mutant neurons from glutamate toxicity as well as astrocytes of WTs [24
]. Studies using mouse models of other neurodegenerative disorders, such as amyotrophic lateral sclerosis [48
], frontotemporal dementia [49
], synucleinopathy [50
], and spinocerebellar ataxia 7 (another polyQ disorder) [51
], provide in vivo
genetic evidence that non-neuronal cells may contribute to neurodegeneration. Thus, converging evidence suggests pathological cell-cell interaction as a common cellular mechanism in multiple neurodegenerative disorders.
Our analyses of the striatal model of HD also indicate that caution should be used when interpreting neurodegenerative phenotypes in HD cellular models over-expressing pathogenic fragments of mhtt. These cellular models have been particularly valuable for rapid dissection of molecular mechanisms underlying the mhtt toxicity. Since neurodegeneration in such models occurs rapidly in days and is due to cell-autonomous toxicity of mhtt, and our striatal model of HD demonstrates that cell-autonomous toxicity of mhtt-exon1 in vivo is insufficient to elicit significant neuropathology, the rapid neurodegenerative process observed in cellular models may reflect different mechanisms than the slow degenerative process which occurs in the in vivo models. The latter may require pathological cell-cell interactions. We suggest that the cellular models are valuable for rapid identification of candidate molecular mechanisms underlying both cell-autonomous and non-cell-autonomous toxicity of mhtt, and these mechanisms should be validated in in vivo genetic models of HD.