This is an interesting paper in that it connects CBP and androgen receptor to the expression of VEGF. (Evidence that CBP regulates VEGF expression and for the connection between CBP and androgen receptor already existed.) It is curious that the authors found reductions only in one transcript of VEGF. If the regulation is at the transcriptional level, one would imagine all alternatively spliced forms to be affected. It would have been better to perform the rescue experiment in vivo, but admittedly, that is a tall order. That said, the authors made a huge effort to avoid problems such as exogenous promoter, overexpression, etc., that are frequently associated with traditional transgenic approaches. These mice do recapitulate the phenotypes of the human disease.

View all comments by Jie Shen

In an elegant study, Sopher et al. generated and analyzed transgenic mice that express yeast artificial chromosomes carrying the human androgen receptor (AR) with either 20 (AR20) or 100 (AR100) CAGs. In this model, human AR is expressed between 30 and 100 percent relative to endogenous mouse AR levels. The AR100 male, but not female, mice developed growth retardation, muscle weakness and atrophy, and motor degeneration resembling X-linked spinal and bulbar muscular atrophy (SBMA, or Kennedy’s disease). The polymorphic CAG repeats in the first exon of AR vary in length from 5-34 in healthy controls to 40-66 in SBMA patients.

The gender-dependent phenotype in AR100 mice is consistent with the requirement of nuclear translocation of mutant AR by testosterone. In transgenic mice expressing AR containing 97 CAG repeats, castration of males rescued the phenotype, while administration of testosterone to the females worsened the manifestations (Katsuno et al., 2002). Based on the requirement for nuclear AR...  Read more

In an elegant study, Sopher et al. generated and analyzed transgenic mice that express yeast artificial chromosomes carrying the human androgen receptor (AR) with either 20 (AR20) or 100 (AR100) CAGs. In this model, human AR is expressed between 30 and 100 percent relative to endogenous mouse AR levels. The AR100 male, but not female, mice developed growth retardation, muscle weakness and atrophy, and motor degeneration resembling X-linked spinal and bulbar muscular atrophy (SBMA, or Kennedy’s disease). The polymorphic CAG repeats in the first exon of AR vary in length from 5-34 in healthy controls to 40-66 in SBMA patients.

The gender-dependent phenotype in AR100 mice is consistent with the requirement of nuclear translocation of mutant AR by testosterone. In transgenic mice expressing AR containing 97 CAG repeats, castration of males rescued the phenotype, while administration of testosterone to the females worsened the manifestations (Katsuno et al., 2002). Based on the requirement for nuclear AR translocation and the fact that AR is a transcription factor, the authors tested whether SBMA transcription is altered by interfering with CREB binding protein (CBP)-mediated transcription of vascular endothelial growth factor (VEGF). CBP is a transcription coactivator that remodels chromatin through its histone acetyltransferase activity. Indeed, coimmunoprecipitation experiments with anti-CBP antibodies revealed more AR in the pulled-down complex with increasing the length of the CAG repeats. In addition, VEGF164, a VEGF isoform expressed in motor neurons and protecting them from apoptotic insults, was reduced in AR100 mice. Finally, adding VEGF164 peptide reduced cell death in a motor neuron cell culture model of AR polyglutamine neurotoxicity.

Strikingly, the affected motor neurons showed diffuse nuclear AR staining but no nuclear aggregates. Nuclear aggregates were observed in unaffected neurons in the dorsal lateral hypothalamus and tectum, in astrocytes in the spinal cord, and in muscle and liver. These findings support the conclusion that nuclear inclusions might not be required for neurotoxicity, and that soluble forms or microaggregates might be toxic. This is also consistent with a mouse model showing a correlation between neuronal vulnerability and solubility of ataxin-1 with 154 glutamines, while nuclear inclusions only occur late in the course of the disease (Watase et al., 2002).

The toxicity of intraneuronal protein aggregates might depend on the neuroprotective potential of the affected cells. While in control mice the herbicide paraquat caused the formation of α-synuclein-containing intraneuronal deposits and generation of nigrostriatal neurons, overexpression of α-synuclein (human wild-type or the Ala53Thr mutant) protected against paraquat-induced neurodegeneration in the presence of the intraneuronal deposits. This neuroprotective effect might be due to increased levels of the neuroprotective HSP70 (Manning-Bog et al., 2003). Overexpression of HSP70 showed neuroprotection without affecting nuclear inclusions in SCA1 mice expressing ataxin-1 with expanded CAG repeats (Cummings et al., 2001) and neuroprotection without affecting α-synuclein containing-intraneuronal deposits in a Drosophila model of Parkinson’s disease (Auluck et al., 2002). Interestingly, overexpression of HSP70 also reduced motor dysfunction in mice expressing AR with 97 CAG repeats (Adachi et al., 2003). Other HSPs such as HSP105α are also neuroprotective against toxicity associated with AR with expanded CAG repeats (Ishihara et al., 2003). The ability of HSPs to reduce toxicity and aggregate formation might be linked or independent. HSP70, HSP40, and HSP27 suppressed toxicity independent of suppression of aggregation (Wyttenbach et al., 2003; Adachi et al., 2003). [Editor's note: see also Magrané et al., 2004.] However, both monomeric AR and nuclear-localized AR complexes were also reported reduced under conditions of overexpression of HSP70, suggesting that HSP70 might be neuroprotective by increasing mutant AR turnover.

View all comments by Jacob Raber