9 March 2011. TDP-43 scouts the nucleus like a roving film editor, splicing out introns from a slew of target RNAs. Researchers believe that identifying those targets will be key in determining how TDP-43 contributes to neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS) and frontotemporal lobar dementia (FTLD). The authors of two papers, posted online February 27 by Nature Neuroscience, used a technique called crosslinking and immunoprecipitation (CLIP) to identify thousands of potential TDP-43 targets, among which hundreds of genes rely on the protein for splicing. “We now have a set of RNAs that we can be really confident are directly regulated by this protein in the brain,” said Jernej Ule of the Medical Research Council (MRC) Laboratory of Molecular Biology in Cambridge, U.K., senior author of one of the studies. The RNAs in this “benchmark dataset,” he said, should be the first that scientists analyze in detail in TDP-43 animal models and cell lines.
Among TDP-43’s most tantalizing targets are the FTLD-related genes FUS and progranulin, TDP-43 itself, the glutamate transporter EAAT2, a bevy of noncoding RNAs, and RNAs harboring extra-long introns that are particularly common in neurons.
Ule worked with joint first authors James Tollervey from his MRC group, Boris Rogelj of King’s College, London, lab of Christopher Shaw, and Tomaz Curk of the University of Ljubljana in Slovenia. The other group, based at the University of California in San Diego (UCSD), consisted of joint first authors Magdalini Polymenidou and Clotilde Lagier-Tourenne in the lab of Don Cleveland, and Kasey Hutt in the lab of Gene Yeo. Alzforum first covered the UCSD team’s efforts at the André-Delambre ALS Symposium held in October in Québec City, Canada (see ARF related news story). Researchers from the University of Texas Southwestern Medical Center in Dallas also published an early TDP-43 target study in November (see ARF related news story on Sephton et al., 2011).
CLIPping TDP-43 RNAs
Each study sought TDP-43-bound RNAs using the CLIP technique. With samples from animals or cell lines, the scientists used ultraviolet light to tightly crosslink nucleic acids and proteins. They then pulled down TDP-43 and any linked RNAs with a TDP-43 antibody. Finally, they sequenced the target codes.
The MRC team used CLIP to compare TDP-43 targets in human postmortem brain tissue taken from three cognitively normal subjects and three people who had sporadic FTLD associated with TDP-43 inclusions. They found some 130,000 sites where TDP-43 could target—not surprising, Ule said, given that TDP-43 binds to UG repeats, the most common small repeat sequence in the human genome. However, he said, TDP-43 may not necessarily modify all the sequences it is able to bind. Removal of approximately 200 exons relied on TDP-43 activity. In those sites, TDP-43 binds broad stretches of RNA, Ule said, suggesting that multiple TDP-43s do so cooperatively.
The researchers found more than 10,000 transcripts that bound TDP-43. They noticed that TDP-43 in FTLD samples was more likely to bind to small, noncoding RNAs, including small nucleolar (sno)RNAs, small nuclear (sn)RNAs, and ribosomal and telomeric RNAs. The samples from healthy controls revealed 2,139 noncoding RNA binding sites, with the FTLD samples having 563 more.
When they examined their data quantitatively, the MRC scientists found that TDP binding to many transcripts was up- or downregulated in the FTLD samples relative to control tissue. Among the differentially binding transcripts were 59 of the most common sequences in the 10,000+ transcripts detected. These included 48 introns, seven 3’ untranslated regions (UTRs), and four noncoding RNAs. Transcripts represented included that for neurexin 3, a protein involved in synaptic plasticity that comes in many splice forms. TDP-43 binding to neurexin 3 mRNA dropped in FTLD tissue compared to normal. Likewise, TDP-43 binding to the transcript for glutamate transporter glial excitatory amino acid transporter-2 (EAAT2), which prevents synaptic excitotoxicity, was also reduced. In FTLD tissue, two noncoding RNAs—nuclear paraspeckle assembly transcript 1 (NEAT1) and metastasis-associated lung adenocarcinoma transcript 1 (MALAT1)—bound more TDP-43.
Noncoding RNAs are part of nuclear speckles and paraspeckles, structures made up of RNAs and splicing factors (reviewed in Lamond and Spector, 2003 and Fox et al., 2002). The study of noncoding RNAs is still an emerging field, said Fen-Biao Gao of the University of Massachusetts Medical School in Worcester, who was not involved with either study. Their function remains unclear, and the data in the study do not indicate whether noncoding RNAs contribute to disease, Ule said.
The UCSD team used CLIP to analyze TDP-43 binding in mouse brains. They discovered nearly 40,000 binding sites in 6,304 genes—some 30 percent of the mouse transcriptome. As in other studies, they confirmed that UG repeats are a popular place for TDP-43 to sit down. However, Yeo noted, there are plenty of UG microsatellites that were not bound to TDP-43, and plenty of places where TDP-43 bound to non-UG sequences, so UG repeats are neither necessary nor sufficient to recruit TDP-43.
Many neuronal and glial transcripts interacted with TDP-43 in the mouse brains, including transcripts for the glutamate transporter Glt1, the myelin-associated glycoprotein Mag, and the myelin oligodendrocyte protein Mog. The UCSD team also found evidence for TDP-43 binding to noncoding RNAs.
The studies complement each other nicely, Ule said. “The binding sites they see are generally in the same positions as we see,” he said. “That is very reassuring.”
Taking TDP-43 Away
TDP-43 can bind to thousands of genes, but which of those candidates really matter? Both groups sought to winnow down their lists to the transcripts most relevant to disease, employing TDP-43 knockout experiments. In people who have ALS or FTLD associated with TDP-43, the protein vacates the nucleus, accumulating and aggregating in the cytoplasm. Thus, the scientists predicted that removing TDP-43 would result in transcriptome changes similar to those in disease.
The MRC group used RNA interference to knock down TDP-43 expression in SH-SY5Y neuroblastoma cells. Using a microarray for 30,154 possible alternatively spliced RNAs, they identified 229 splicing changes in the TDP-43-free cultures. Many were transcripts encoding neural genes. TDP-43 also promoted alternative splicing of Bcl-2 interacting mediator of cell death, or BIM, suppressing the gene’s most toxic isoform. In subjects with FTLD, levels of the toxic isoform were higher than normal, so BIM might contribute to neurodegeneration, the authors wrote.
The UCSD team used antisense oligonucleotides to block TDP-43 translation in the striatum of mouse brains. Control mice received an antisense oligo unrelated to any sequence in their genome. Two weeks later, the scientists sacrificed the animals, dissected out the striatum, and sequenced the RNAs.
With these data in hand, the scientists discovered that the drop in TDP-43 upregulated 362 genes and downregulated 239 more. Noncoding RNAs were also affected, with four increased and 55—including MALAT1—decreased in the TDP-43 depleted brains. Affected genes included FUS and progranulin, both linked to FTLD. The researchers also detected 965 splice form alterations in the TDP-43 knockdown samples.
Among the genes downregulated in the absence of TDP-43, many contained very long introns with multiple sites for TDP-43 to bind. Indeed, the most downregulated genes contained introns an average of 28,707 base pairs long, compared to an average of 4,532 base pairs for unaffected or upregulated genes. Many TDP-43-regulated, long-intron genes had neural functions, including neurexins 1 and 3. The authors suggested that the long introns contain regulatory elements.
Genes expressed in the central nervous system tend to have long introns, Yeo said. Thus, the data provide a potential explanation for why neurons alone are vulnerable to mutations in TDP-43, even though the gene is ubiquitously expressed. Yeo said: “The targets are key to neuronal vulnerability.” Neurons need these particular genes with long RNAs, and they need TDP-43 to protect those transcripts, he speculated.
TDP-43 Turning on Itself
Scientists have previously shown that TDP-43 regulates its own transcript (see ARF related news story on Igaz et al., 2011); the UCSD group figured out a mechanism. They discovered that the protein TDP-43 binds to TDP-43 mRNA at the site of an alternatively spliced intron in the TDP-43 3’ UTR. Splicing out that intron tags the TDP-43 transcript for nonsense-mediated decay.
Nonsense-mediated decay is based on the proximity of stop codons and exon-exon junctions. When a splicing factor such as TDP-43 connects two exons, it leaves behind a protein complex that travels to the ribosome along with the transcript. As long as the stop codon is close to the final exon-exon junction, the ribosome can “assume” that it is an appropriate stop codon and it knocks off the exon-exon protein complex to translate the transcript. But if the stop codon comes more than 50 nucleotides upstream of another exon-exon junction, the ribosome “assumes” that the stop codon is a nonsense mutation. This instigates a series of reactions leading to uncapping and degradation of the faulty mRNA. Therefore, by splicing out an intron in the 3’ UTR, TDP-43 can prevent a transcript from being translated.
Thus, if there is plenty of TDP-43 in the nucleus, it will splice the 3’ UTR of its own transcripts, preventing the cell from producing more of the protein. But if TDP-43 is out in the cytoplasm—as happens in disease—then TDP-43 transcripts in the nucleus would go unmodified. They would be able to produce protein in the ribosome and lead to ever-increasing TDP-43 protein concentrations. “This may be why you get accumulation of more and more TDP-43 in the cytoplasm of patients,” Yeo said.
This kind of combination of biology and bioinformatics provides crucial insights, Yeo said. “There will be an avalanche of data coming through.” He and Cleveland have already started another collaboration: They are repeating their CLIP experiments with FUS, another RNA partner protein linked to FTLD.—Amber Dance.
Tollervey JR, Curk T, Rogelj B, Briese M, Cereda M, Kayikci M, König J, Hortobágyi T, Nishimura AL, Zupunski V, Patani R, Chandran S, Rot G, Zupan B, Shaw CE, Ule J. Characterizing the RNA targets and position-dependent splicing regulation by TDP-43. Nat Neurosci. 2011 Feb 27. Abstract
Polymenidou M, Lagier-Tourenne C, Hutt KR, Huelga SC, Moran J, Liang TY, Ling SC, Sun E, Wancewicz E, Mazur C, Kordasiewicz H, Sedaghat Y, Donohue JP, Shiue L, Bennett CF, Yeo GW, Cleveland DW. Long pre-mRNA depletion and RNA missplicing contribute to neuronal vulnerability from loss of TDP-43. Nat Neurosci. 2011 Feb 27. Abstract