10 May 2011. TDP-43 and FUS: so alike. Both the TAR DNA binding protein-43 and the fused in sarcoma genes, when mutated, cause amyotrophic lateral sclerosis. Both encode RNA-binding proteins with prion-like domains and, during disease, they exit the nucleus and form nasty aggregates in the cytoplasm. Small wonder, then, that scientists expected they might damage cells in the same manner. Not so, according to two papers in the April PLoS Biology. Using yeast models, Aaron Gitler and colleagues at the University of Pennsylvania School of Medicine found crucial differences between how FUS and TDP-43 aggregate, and how they affect the cell downstream. In the same journal, scientists led by Gregory Petsko at Brandeis University in Waltham, Massachusetts, report complementary results from their own yeast model for FUS toxicity, showing some of the same FUS domains and other genes involved in the protein’s killer activities as the Pennsylvania team. And if multicellular organisms are more to your liking, consider the Drosophila FUS model published online in Human Molecular Genetics April 12. The authors found that FUS and TDP-43 mutants interact synergistically to cause neurodegeneration, suggesting their distinct pathways do intersect at some point in the toxicity process.
Yeast models are definitely in vogue. Gitler made a TDP-43 yeast model some years ago, and showed that, when overexpressed, the protein moves into the cytoplasm and gloms together, leading to a toxic phenotype (see ARF related news story on Johnson et al., 2008). More recently, the lab pinned an ataxin-2 homolog as an enhancer of TDP-43 toxicity in yeast, leading the researchers to discover that ataxin-2 polyglutamine expansions are linked to ALS in people (see ARF related news story on Elden et al., 2010). Two other yeast models for FUS were also published recently (Kryndushkin et al., 2011; Fushimi et al., 2011).
Gitler teamed with joint first authors Zhihui Sun and Zamia Diaz and co-senior author James Shorter for a repeat performance with a yeast model for FUS. At the same time, Petsko, first author Shulin Ju, and joint senior author Dagmar Ringe were making their own version. “From the start, we were kind of expecting [TDP-43 and FUS] to behave very similarly,” Shorter said. “We were surprised when many differences turned up.” FUS aggregation requires an extra domain, in addition to the RNA-binding and prion-like domain that TDP-43 needs to form inclusions. While ALS-linked TDP-43 mutations affect the protein’s toxicity, the common FUS mutations only influence its location. Plus, in suppressor and enhancer screens, the researchers found that the genes that mediate each protein’s toxicity are very different sets.
Not surprisingly, mutant versions of FUS formed cytoplasmic inclusions in yeast, and high doses of the mutant protein were deadly. But wild-type FUS, the researchers found, behaved the same way. “Both turned out to be toxic,” Petsko said, concluding, “disruption of FUS localization is sufficient for toxicity.” Thus, the groups concentrated on wild-type protein in further experiments.
FUS contains a non-classical nuclear localization sequence (NLS) at its carboxyl terminus, where most ALS-linked mutations in FUS are; however, in yeast, FUS was primarily in the cytoplasm. Ju, in Petsko’s lab, confirmed that FUS’s unconventional NLS does not work in yeast cells, making it a good model for cytoplasmic gain of toxic function, but a poor model for nuclear effects of FUS mutations. When the researchers artificially routed FUS to the nucleus, it failed to aggregate, or kill cells, confirming that the cytoplasmic location is key for aggregation and toxicity. These data confirm that an early event in FUSopathy is the protein’s mislocalization (see ARF related news story on Dormann et al., 2010). In TDP-43 proteinopathy, too, cytoplasmic mislocalization of the protein is a key event.
FUS and TDP-43 are similar in structure: Both contain RNA recognition motifs (RRMs), prion-like domains, and glycine-rich segments. In TDP-43, the RRMs and the carboxyl terminal prion domains are involved in aggregation; most ALS-linked mutations are in this region. What about FUS? Sun and colleagues created truncated versions of the protein and examined the localization and toxicity of each. Like TDP-43, FUS requires the RRM and prion-like segments to aggregate, but also a region rich in arginine-glycine-glycine (RGG) repeats, which are not found in TDP-43. Diaz and Shorter confirmed the RGG requirement in in vitro experiments. However, removing the carboxyl terminal NLS of FUS—the site of ALS mutations—did not prevent aggregation, further confirming that those mutations do not promote aggregation directly. They simply alter the protein’s location. In contrast, TDP-43 mutations seem to affect mainly the aggregation pathway.
The researchers also examined which parts of the proteins contributed to toxicity. In TDP-43, those were the same regions contributing to aggregation, the RRMs, and the carboxyl terminal prion domains. In FUS, the RRM, the prion-like domain, and an RGG region were necessary to kill yeast, but the carboxyl terminus was unnecessary. In fact, both groups found that FUS was more toxic when the carboxyl terminus was missing, matching reports that people with nonsense mutations in that region get sick earlier and progress more rapidly (Waibel et al., 2010).
Accessories to Toxicity
Having identified the parts of FUS responsible for its toxicity, both teams set out to discover what other happenings in the cell might intensify or alleviate that toxicity. Each team screened some 5,500 overexpressed yeast genes. Petsko and colleagues looked for suppressors, while Gitler and colleagues sought both suppressors and enhancers. The Petsko team identified five suppressors. Gitler’s group picked out those five as well, plus 19 more, and 10 enhancers. The five overlapping suppressors were translation terminator ECM32, RNA binding protein NAM8, snRNP associate SBP1, transcription factor SKO1, and transcriptional activator VHR1.
Do any of these also modulate the toxicity of TDP-43? In unpublished work, Gitler and colleagues identified 40 yeast genes that do, but only two of those overlapped with the FUS list. Thus, the proteins not only aggregate differently, but they also cause cell death by different pathways.
Genetic screens frequently yield up to 100 or more hits, leaving Petsko surprised at his modest catch. “Such a small number of genes were found, and they were all RNA binding proteins,” he said. In line with what many other studies have suggested, Petsko concluded that “RNA homeostasis lies at the heart of FUS toxicity.”
Among the Gitler group’s hits were suppressors Pab1 and LSM7. Both participate in assembling stress granules: clusters of RNAs and RNA binding proteins that temporarily form to protect cells from stress (see ARF Webinar). Gitler said that this is the first direct evidence that FUS not only goes to stress granules, but also that the granules are part of its toxic modus operandi. Those granules might trap necessary RNAs, and their loss could contribute to toxicity, suggested Richard Gardner of the University of Washington in Seattle, who was not involved in the current works.
Curiously, the genes that suppressed FUS toxicity did so without dissolving FUS aggregates. That is good news for therapy, Petsko said. While he could not imagine how a drug could bust the globs and send mutant FUS back to the nucleus, he easily envisioned a small molecule acting in downstream toxicity. Another reason to be optimistic about future therapies is that the Petsko team showed that the human homolog of one of the FUS suppressors, hUPF1, also ameliorated toxicity in yeast. This suggests that the human FUS-response system mirrors the yeast system in at least some respects. The yeast FUS models would be suitable for screening potential drugs.
Separate, But Intersecting
The study authors conclude that TDP-43 and FUS are not interchangeable killers. They use different domains to form inclusions, and interact with different genes. “Yes, both proteins misfold, but they are probably engaging quite different downstream pathways to become toxic to the cells,” said Paul Muchowski of the Gladstone Institute at the University of California, San Francisco, who was not involved with either study.
“Pathogenic differences between FUS and TDP-43 argue compellingly that ALS is not a single disease, but rather a spectrum of clinically convergent but mechanistically disparate disorders,” wrote Lary Walker of Emory University in Atlanta, Georgia, in an e-mail to ARF. “Hence, for both practical and theoretical reasons, it makes sense to define these variants according to the main proteins involved in their pathogenesis,” added Walker. He was not involved in either study.
For all their differences, the new fruit fly model suggests that the TDP-43 and FUS pathways are not wholly separate. First author Nicholas Lanson, Jr., and senior author Udai Pandey of the Louisiana State University Health Sciences Center in New Orleans, and colleagues created flies that carry human FUS genes. FUS mutants caused neurodegeneration and defects in locomotion. The nuclear localization signal worked in the flies, so that wild-type FUS was nuclear, but mutants were cytoplasmic. Corroborating the yeast work, the scientists also found that forcing FUS into the nucleus was not toxic. In addition, the researchers co-overexpressed wild-type FUS and mutant TDP-43, and found the two together wrought four times the damage to the fly eye than either alone. This synergistic interaction suggests that the pathways by which each causes toxicity may intersect at some point.—Amber Dance.
Sun Z, Diaz Z, Fang X, Hart MP, Chesi A, Shorter J, Gitler AD. Molecular determinants and genetic modifiers of aggregation and toxicity for the ALS disease protein FUS/TLS. PLoS Biol. 2011 Apr;9(4):e1000614. Abstract
Ju S, Tardiff DF, Han H, Divya K, Zhong Q, Maquat LE, Bosco DA, Hayward LJ, Brown RH Jr, Lindquist S, Ringe D, Petsko GA. A yeast model of FUS/TLS-dependent cytotoxicity. PLoS Biol. 2011 Apr;9(4):e1001052. Abstract
Lanson NA Jr, Maltare A, King H, Smith R, Kim JH, Taylor JP, Lloyd TE, Pandey UB. A Drosophila model of FUS-related neurodegeneration reveals genetic interaction between FUS and TDP-43. Hum Mol Genet. 2011 Apr 12. Abstract