G quadruplex RNA structures have been shown, for the first time, to form in tissue from C9ORF72 ALS patients. There they trap the splicing factor hnRNP H, leading to missplicing of its targets in patient brain, according to a new study in the September 13 eLife. The study strongly implicates hnRNP H sequestration as a pathogenic event in C9ORF72 ALS, according to first author Erin Conlon and principal investigator James Manley of Columbia University.
“I think this makes it the leading candidate for the protein that interacts in an important way with the repeats,” Manley said.
RNA-based G quadruplexes (G-Qs) form when guanine-rich RNA folds back on itself, allowing guanine-guanine hydrogen bonding to hold the RNA in a rigid, three-dimensional structure (Payet and Huppert 2012). RNA G-Qs have been proposed to form from the expanded hexanucleotide repeats in C9ORF72 ALS (see Fratta et al., 2012, Reddy et al., 2013, Nov 2013 news), but whether they are common in the C9 ALS/FTD brain, and what their contribution to pathogenesis might be, is unknown, Manley said.
In order to investigate how C9 repeat RNA may contribute to neuron death, the team first set out to identify which proteins bind with these repeats and how the interactions are affected by RNA conformation. In vitro, they combined a brain cell-derived nuclear extract with RNA containing 10 C9 repeats, a length known to be able to form G-Qs. When they UV crosslinked the RNA-protein complexes, they detected a previously unknown protein interaction with the RNA. Through multiple techniques, including use of the G-Q-specific BG4 antibody, immunoprecipitation, and siRNA depletion, they showed that the RNA formed G-Qs, and that heterogeneous ribonucleoprotein H (hnRNP H) was, the authors wrote,the “predominant, and perhaps only” protein crosslinking the RNA in these cells.
They next sought to determine if G-Qs formed and bound hnRNP H in patient-derived cells, using BG4 to visualize them in fibroblasts from seven C9 patients. They found that patient-derived cells contained 1.8 times as many G-Q-positive foci as control cells. Staining in astrocytes from autopsy of a single C9 patient showed a similar increase compared to control astrocytes. When the authors stained for both G-Qs and hnRNP H, foci containing the pair were more than three times as common in C9 astrocytes versus control astrocytes. In two C9 patient spinal cords, they counted 46 large hnRNP H-positive foci in 36 motor neurons, versus three in 19 motor neurons from two control spinal cords.
The team wanted to understand whether the hnRNP H-G-Q foci they observed sequestered significant amounts of the splicing factor, such that it might affect gene expression. They quantified the amount of hnRNP H in brain tissue extracts that was insoluble, and thus presumably sequestered in RNA foci. They found that the average percentage of insoluble hnRNP H was 30% in non-ALS motor cortex, versus 56% in C9 ALS (p=0.013), a difference they suggested is likely to be biologically relevant.
Does binding of hnRNP to G-Qs alter splicing in ALS brain? To answer that, the authors analyzed alternative splicing in seven C9 ALS, one SOD1 ALS, and six non-ALS control cerebellums, chosen because of tissue availability, RNA quality, and the recognized cerebellar atrophy in C9 disease. They chose 13 exon targets known to be affected by hnRNP H and linked to neurologic disease or disease-related pathways, as well as three control exons known to targeted by a separate splicing factor, SRSF2, but with no known link to hnRNP H.
Looking first in U87 glioblastoma cells to validate the presumed effect of hnRNP H on these exons, they found that 12 of the 13 hnRNP H targets, and two of the three SRSF2 targets, were affected by hnRNP H knockdown. In cerebellar C9 ALS tissue, they found that 11 of the 12 hnRNP H targets, and one of the two SFRF2 targets, showed a change in the same direction as seen in the knockdown experiments, as compared to non ALS or SOD1 ALS tissue. “These changes are consistent with a loss of function of hnRNP H in C9ORF72 patient brain,” Conlon said.
G quadruplexes are likely to be stabilized by the presence of hnRNP H, Conlon said, though the mechanism remains unknown. “It’s not necessarily the only mechanism of increased stability, but we think it is an important part of why these aggregates form.” That stability then likely enhances aggregation, Manley added, making the complexes resistant to intracellular nucleases that would otherwise degrade them. This would allow the aggregates to continue acting as hnRNP H sinks over long periods of time. This suggests that in C9-based ALS, the production of G quadruplexes over a lifetime might lead to ever-greater sequestration of hnRNP H, altering expression of many genes.
“A number of papers have investigated proteins that associate with the GGGGCC repeat,” commented Benjamin Wolozin of Boston University. “Where this paper stands out is in its breadth, from the demonstration of binding to G quadruplexes all the way up to splicing changes in patient tissue.”
However, he cautioned, “There isn’t any evidence currently that the association of anything with RNA foci is necessarily deleterious or the cause of degeneration,” noting that foci appear in some brain regions where there appears to be little cell loss. “Showing that something is associated with repeats is different than proving it contributes to degeneration”, he pointed out.
“Therefore, the question remains of how important this [hnRNP H-G-Q binding] is for the disease process,” he said, given that the C9 repeat RNA probably binds other proteins, and there may be non-foci related pathogenic mechanisms at work as well. “Nonetheless, to demonstrate splicing changes related to loss of hnRNP H is impressive, and a big plus for this work.”
Conlon EG, Lu L, Sharma A, Yamazaki T, Tang T, Shneider NA, Manley JL. The C9ORF72 GGGGCC expansion forms RNA G-quadruplex inclusions and sequesters hnRNP H to disrupt splicing in ALS brains. eLife. 2016 Sep 13;5. pii: e17820. [Pubmed].