24 May 2011. They walk clumsily, slur their speech, and show similar brain pathology as people with Huntington’s disease (HD). And, like HD sufferers, these rare individuals with Huntington’s disease-like 2 (HDL2) have the same type of genetic mutation—an expanse of DNA encoding abnormally long glutamine stretches. But whereas the culprit gene in HD is huntingtin, in HDL2 it is junctophilin-3 (JPH3). How can mutations in such different genes cause the same symptoms? Modeling HDL2 in a series of transgenic mice that debuted in the May 12 Neuron, researchers led by X. William Yang at the University of California, Los Angeles, clarify the disease’s molecular underpinnings—namely, that an antisense CAG transcript driven by a novel promoter can trigger HDL2 pathogenesis, and that it may do so in ways similar to those in other polyglutamine diseases, including HD.
Beyond the symptoms (Margolis et al., 2001) and neuropathological features (Greenstein et al., 2007; Rudnicki et al., 2008) it shares with HD, HDL2 has been a tougher nut to crack. While HD clearly arises from long CAG repeats giving rise to polyglutamine(polyQ)-expanded huntingtin proteins, the JPH3 gene includes leucine-encoding CTG repeats in an alternatively spliced exon (Holmes et al., 2001). The antisense strand could produce CAG-repeat RNAs that code for polyglutamine, but there is no evidence to date for such transcripts in HDL2 patients. And if they do exist, are they translated? Meanwhile, scientists have detected clusters of the sense CUG transcripts in postmortem HDL2 brain tissue, and showed that these aggregates can trap other proteins and keep them from doing their jobs (Rudnicki et al., 2007). Furthermore, HDL2 brains have HD-like nuclear inclusions that stain with polyQ antibodies, suggesting the lesions contain polyQ peptides. However, the polyQ nuclear bodies rarely colocalize with the CUG RNA clusters, suggesting these pathological elements might drive disease in separate ways.
As a step toward unraveling these issues, the UCLA team used a bacterial artificial chromosome (BAC) to make transgenic mice carrying the entire 95 kb human JPH3 gene, including approximately 120 CTG repeats. Healthy people have six to 28 CTG repeats in JPH3, while HDL2 patients carry 40 to 59. The authors designed the BAC mice with extra-long repeats to speed the disease process.
By and large, first author Brian Wilburn and colleagues succeeded in capturing the human disease in these transgenics. Unlike their wild-type littermates, BAC-HDL2 mice developed motor deficits and lost forebrain neurons. In addition, their brains accumulated the two pathological hallmarks, polyQ-immunoreactive inclusion bodies and CUG RNA foci, with structure and distribution resembling HDL2. All these features intensified as the mice aged.
To make sure the neuropathology derived specifically from the long string of CTGs, and not simply from the presence of a BAC transgene, the researchers created two BAC control strains carrying the human JPH3 locus with just 14 CTG repeats. One had the same background (FvB/N) as the BAC-HDL2 mice; the other was made on a C57/BL6 background. Neither BAC control line developed motor deficits or neuropathological signs of disease.
Do these CTG/CAG stretches in fact give rise to the polyQ proteins found in HDL2 brains? In the absence of proof for polyQ-encoding CAG transcripts in human samples, Wilburn and colleagues produced the next best thing—circumstantial evidence in cells and mice. Using reverse-transcriptase PCR, they detected CAG transcripts from the JPH3 antisense strand in BAC-HDL2 mice, and with a bit of in silico investigation, found a putative promoter preceding the polyQ open reading frame on the antisense strand of human JPH3. They showed that genomic fragments encompassing this region could drive expression of a luciferase reporter in primary cortical neurons. Furthermore, they showed that BAC-HDL2 brain extracts contain a polyQ protein around the same size as 120-repeat HDL2-CAG protein from transfected HEK293 cells. These experiments provide evidence for a novel promoter that drives expression of a polyQ-encoding RNA made when JPH3 is transcribed in the antisense direction.
The UCLA researchers then demonstrated that these polyQ-encoding transcripts are sufficient to drive pathogenesis in mice. They created another transgenic strain, BAC-HDL2-STOP, in which sense transcription of the JPH3 transgene is blocked by a STOP sequence inserted into exon 1. As such, BAC-HDL2-STOP mice do not make CUG RNAs or JPH3 protein, but generate CAG transcripts from the antisense JPH3 strand as usual. Importantly, these mice develop motor impairment and polyQ nuclear inclusions as they age, showing that CAG antisense transcripts by themselves can, in fact, trigger HDL2 pathogenesis.
Further experiments suggested that HDL2’s resemblance to HD may run even deeper. The scientists detected the transcriptional coactivator CBP in the polyQ nuclear bodies from BAC-HDL2, BAC-HDL2-STOP, as well as human HDL2 brains, and this sequestering led to reduced expression of a CBP target gene, brain-derived neurotrophic factor (BDNF). Transcriptional downregulation of BDNF, a critical growth factor for striatal neurons, has been implicated in HD pathogenesis (reviewed in Zuccato et al., 2010).
All told, the current study “demonstrates a very elegant murine genetic approach for ascertaining the biological impact of an antisense CAG transcript and provides support for HDL2 being a polyQ disease,” wrote Harry Orr of the University of Minnesota, Minneapolis, in an editorial accompanying the paper. He noted that the work does not exclude the possibility that a toxic RNA from the sense CUG strand could also contribute to disease.
Christopher Ross of Johns Hopkins University, Baltimore, Maryland, said the convergent disease mechanisms uncovered in the BAC-HDL2 mice will “give us good clues as to what is important for HD pathogenesis.” A recent PNAS paper also examines HD’s underpinnings, albeit at the cellular level. It describes neuronal network changes underlying the cognitive slide that can precede classic motor symptoms in HD (see ARF related news story).
On a broader level, the Neuron report suggests the need to examine antisense repeat-containing transcripts in other brain diseases. “Bidirectional transcription is coming to the fore as an important biologic and potentially disease-causing mechanism,” Ross said. Other trinucleotide repeat disorders including spinocerebellar ataxia type 8 (SCA8) and myotonic dystrophy type 1 (DM1) also have such bidirectional transcription, Yang told ARF. Preliminary data from other labs suggest the mechanism operates in SCA7 and HD as well.—Esther Landhuis.
Wilburn B, Rudnicki DD, Zhao J, Weitz TM, Cheng Y, Gu X, Greiner E, Park CS, Wang N, Sopher BL, La Spada AR, Osmand A, Margolis RL, Sun YE, Yang XW. An antisense CAG repeat transcript at JPH3 locus mediates expanded polyglutamine protein toxicity in Huntington’s Disease-like 2 mice. Neuron. 12 May 2011;70:427-440. Abstract
Orr HT. Are polyglutamine diseases expanding? Neuron. 12 May 2011;70:377-378. Abstract