17 July 2009. Researchers have identified a guardian angel protein that protects neurons from the toxic effects of nitric oxide. In yesterday’s issue of Neuron, scientists from the Johns Hopkins University School of Medicine in Baltimore, Maryland, describe the role of GOSPEL (GAPDH’s competitor of Siah Protein Enhances Life) in preventing GAPDH-mediated cytotoxicity. When modified by nitric oxide, GOSPEL keeps GAPDH in the cytosol, preventing it from moving to the nucleus where it can induce apoptosis. GOSPEL protected cells, both in culture and in the brains of mice, from a neurotoxin, although no direct role in preventing or delaying neurodegenerative disease has yet been shown. The work of joint first authors Nilkantha Sen and Makoto Hara, principal investigators Akira Sawa and Solomon Snyder, and colleagues was 12 years in the making; Sawa first pulled a GOSPEL fragment out of a yeast two-hybrid screen while still a postdoc in Snyder’s lab before starting his own lab.
“I think most people know [GAPDH] as a boring glycolytic protein; you use it as a loading control,” said Robert Cumming of the University of Western Ontario in London. Cumming, who studies GAPDH, was not involved in the current study. But Sawa’s and Snyder’s labs have shown it also acts as a cell stress sensor, and can be a killer when the conditions are right. Nitric oxide (NO) synthases, activated by cell stress, can modify GAPDH with an NO attachment to one of its cysteines in a process called S-nitrosylation (Hara et al., 2005). Once nitrosylated, GAPDH can interact with Siah, a ubiquitin ligase with a nuclear localization sequence. Siah drags GAPDH into the nucleus, where it helps to stabilize Siah. There the pair degrade Siah’s targets and activate CREB binding protein, which in turn stimulates downstream apoptotic proteins such as p53 (Sen et al., 2008).
“This kind of fits into a bigger picture of nitrosylation,” said Stuart Lipton of the Burnham Institute in La Jolla, California, who co-wrote a preview accompanying the Neuron paper. Nitrosylation will prove to be an important protein modifier, he said: “It’s like phosphorylation was 40 years ago.”
Nitric oxide and GAPDH are already known to mediate cytotoxicity in neurodegenerative disorders. In Alzheimer disease, S-nitrosylation of a mitochondrial fission protein in the presence of amyloid-β contributes to neuronal death (see ARF related news story and Cho et al., 2009), and the same modification of parkin may be linked to Parkinson disease (see ARF related news story and Chung et al., 2004). In Huntington disease, binding of mutant huntingtin to GAPDH and Siah facilitates the protein’s translocation to the nucleus (Bae et al., 2006).
To find proteins that interacted with GAPDH, in 1997 Sawa and associates performed a yeast two-hybrid screen with the carboxyl- and amino-termini of the protein. The carboxyl terminus bound to Siah, the protein he chose to focus on first because its function in cellular toxicity was already known. After publishing the Siah work in 2005, Sawa and colleagues turned to another fragment that had bound the amino terminus of GAPDH—the fragment that turned out to be GOSPEL. If Siah is the devil on GAPDH’s left shoulder, GOSPEL is the angel on the right.
GOSPEL, a cytosolic protein, is highly expressed in brain, heart, lung, and skeletal muscle; GAPDH is ubiquitous and present at high levels in the same tissues. In brain, GOSPEL is strongly expressed in neurons, particularly those that contain plenty of GAPDH such as cerebellar Purkinje cells, CA1-3 pyramidal cells, and hippocampal granule cells in the dentate gyrus.
The researchers found that GOSPEL, like GAPDH, can be S-nitrosylated. Treating cerebral cortical cells with NMDA, which activates nitric oxide synthases, resulted in S-nitrosylation of GOSPEL; this modification was abolished in a mutant lacking the protein’s single cysteine. GOSPEL and GAPDH could bind without the nitrosylation, but it greatly enhanced their interaction in cortical neurons as well as HEK293 cells. GAPDH S-nitrosylation also improved its affinity for GOSPEL. Similarly, GAPDH-Siah binding is augmented when GAPDH is S-nitrosylated. Sawa has unpublished data that other oxidative forms of GAPDH may be even more important to trigger the GAPDH-Siah signaling cascade.
Although they interact with opposite ends of GAPDH, Siah and GOSPEL compete to bind it. Overexpression of GOSPEL, in HEK293 cells treated with a nitric oxide donor, diminished GAPDH-Siah binding. Sawa does not yet know how the binding of GOSPEL at one end of GAPDH prevents Siah from binding at the other (and vice versa), but suspects allosteric interactions are involved. It is also unclear what determines whether Siah or GOSPEL will hook up with GAPDH, but time may be a factor. Nitrosylated GOSPEL is detectable three hours after NMDA treatment, but GAPDH nitrosylation does not occur until six hours in. This time difference may help cells that are only mildly stressed maintain homeostasis, Sawa theorized, because the initial response to stress is for GOSPEL to bind GAPDH, protecting the cell. But if the stress continues, GAPDH also gets nitrosylated and is able to bind Siah and induce cytotoxicity. “Once the stress is over the threshold, the cell decides to die,” Sawa suggested.
To determine GOSPEL’s mode of action, the researchers used RNAi to deplete the protein from primary cerebral cortex cultures. In GOSPEL-depleted cells treated with NMDA, the GAPDH-Siah pairing was increased over that in control cells—further evidence that GOSPEL normally inhibits the interaction. The GOSPEL RNAi also increased the levels of GAPDH in the nucleus. Then the scientists overexpressed GOSPEL in HEK293 cells. Normally treating these cells with the nitric oxide donor GSNO causes GAPDH to move to the nucleus, but GOSPEL diminished nuclear GAPDH. Therefore, GOSPEL’s binding keeps GAPDH in the cytosol where it cannot cause trouble.
Next, Sawa and colleagues looked for an effect of GOSPEL on neurotoxicity. Treating primary cerebellar granule neurons with NMDA normally kills 75 percent of the cells. But the majority of cells overexpressing GOSPEL survived. A GOSPEL mutant lacking the S-nitrosylation site was not neuroprotective, confirming the role of nitrosylation in its activity. Similarly, a GOSPEL fragment lacking the GAPDH-binding segment failed to improve cell survival, showing its interaction with GAPDH is essential for neuroprotection. When the researchers used RNAi to deplete endogenous GOSPEL in primary neural cultures, they found that NMDA toxicity was increased, with fewer than half as many cells surviving as in control cultures.
GOSPEL, then, can protect cells from excitotoxicity, but the researchers wondered if it would do the same in animals. They used a viral vector to deliver two forms of GOSPEL to the cerebral cortex of mice: wild-type protein and a deletion mutant missing the region that binds to GAPDH. Then they injected NMDA into the cerebral cortex. The lesions in the animals overexpressing wild-type GOSPEL were 30 percent smaller than those with the mutant, confirming that GOSPEL-GAPDH binding is required to protect neurons.
The next step will be for other scientists to confirm the role of GAPDH and GOSPEL in other models. “The data are fairly good,” Cumming said, but “I’d like to see some other labs publish similar stuff…it would be nice to see it in some other cell models.” Lipton would like to know if neuron-specific depletion of GOSPEL makes an animal more susceptible to nitric oxide toxicity; it would then be interesting to cross such a mouse to other models of neurodegenerative disease. “It is a beautiful, elegant paper, but a bit of it is inferential,” he said.
Sawa is excited about the clinical implications of the work. “I am very optimistic that some compound from this cascade will have some benefit for the many diseases in which GAPDH is involved,” he said. There is already a drug, deprenyl, which prevents GAPDH-Siah binding (Hara et al., 2006). “The deprenyl binding domain may be very similar to the GOSPEL binding domain,” Sawa said, in which case the drug might inhibit Siah binding by the same, perhaps allosteric, mechanism that GOSPEL does. Deprenyl, which is also an MAOI inhibitor, is used to treat symptoms of Parkinson’s. But any drug targeted at the GOSPEL-GAPDH interaction, Sawa said, would have to be specific to that interaction.
“If you want to target GAPDH, you have to be careful,” Cumming said, because of the protein’s essential role in metabolism. If GOSPEL’s neuroprotective role is confirmed, he said, then augmenting GOSPEL activity might be a safer target: “This is a likely alternative to go after in terms of manipulating GAPDH’s ‘darker side.’”—Amber Dance.
Sen N, Hara MR, Ahmad AS, Cascio MB, Kamiya A, Ehmsen JT, Aggrawal N, Hester L, Doré S, Snyder SH, Sawa A. GOSPEL: A neuroprotective protein that binds to GAPDH upton S-nitrosylation. Neuron. 2009 Jul 16;63:81-91. Abstract
Nakamura T, Lipton SA. According to GOSPEL: Filling in the GAP(DH) or NO-mediated neurotoxicity. Neuron. 2009 Jul 16;63:3-6. Abstract