For gene chasers, surprises are common and part of the adventure of scientific discovery. Sometimes, we can define a new function for a previously unknown protein by identifying the phenotype when that protein is missing. Other times, we stumble into a very mature field, where our discovery is welcomed, and we have to catch up on many years of research. That was the case when we recently discovered mutations in the ATG5 gene encoding a core macroautophagy (hereafter autophagy) protein in 2 subjects with ataxia.
In 2005, we recruited a family with 2 subjects with recessive ataxia from Turkey based on their clinical description being similar to Cayman ataxia, a congenital form of ataxia whose gene we had identified in 2003. The affected individuals had early delayed milestones such as late walking, difficulties articulating, as well as lower IQ. MRI scans showed a smaller cerebellum. However, in contrast to the most common forms of ataxia, the disease is not progressive. While the subjects are neurologically and cognitively impaired and live with their parents, they can walk unaided, hold jobs and report even slight improvements in symptoms over the years.
We quickly found that the gene involved is different from that of Cayman ataxia, but could not do much more at the time. Ten years later, exome sequencing opened new avenues for families with rare disorders. But exome sequencing usually identifies far too many potentially damaging mutations in single cases or even siblings. Here, we were helped by the fact that the parents were from the same relatively remote village, and hence we postulated that a single identical region of the genome was transmitted from a common ancestor. Only a single large stretch on chromosome 6 was inherited in a homozygous (identical) state in common between the 2 patients but not the 2 unaffected siblings. In that interval, we found only one homozygous and computationally predicted to be damaging mutation, an E122D change in ATG5. An E/Glu to D/Asp mutation can be subtle, as it does not change the charge of the amino acid. In this case, however, the degree of conservation was significant and led to the bioinformatic prediction: Glu122 is conserved in all ATG5 proteins from various eukaryotic species, even down to yeast! So, here began our exploration of autophagy, helped by lymphoblastoid cell lines (LCLs) we had generated from blood from all patients, and by many colleagues in many institutions and countries.
We first investigated the effect of the E122D (E141D in yeast) mutation in a yeast model, which is proven to be an excellent system for genetic studies of autophagy. Yeast data demonstrated that the mutation can reduce autophagic flux by 30–50%, measured by diverse methods. This result suggested that the mutant Atg5/ATG5 is only partially active, a “hypomorph.”
In both yeast and mammals, we found the mutation to be located in the ATG5 structure close to the site of ATG12–ATG5 conjugation. In fact, in a blinded experiment, comparing ATG5-mutated patient LCLs to those of other patients or controls, expression of the ATG12–ATG5 conjugate was dramatically reduced by the ATG5E122D mutation. E122D-mutated ATG5 was also found to be defective in conjugating with ATG12, when heterologously expressed in HEK293 cells, Hi5 insect cells and Drosophila tissues. In ATG5E122D patient LCLs, autophagic flux was also prominently attenuated, monitored by time-dependent LC3 lipidation (a process in which ATG12–ATG5 acts as an E3 ligase).
Collaborators in Hungary already had generated flies missing Atg5. After receiving these flies, which are alive but very sick, we could check complementation with wild-type and mutant forms of human ATG5. Quantifying movement disorder turned out to be a challenge, but as shown in the movies in the publication, the results were clear: First, amazingly, human ATG5 was able to substantially complement the Atg5-null fly phenotypes, illustrating how conserved the autophagy system is across species. Second, the E122D mutant ATG5 was better than the vector control, but worse than wild-type human ATG5, indicating that E122D is clearly a “hypomorph,” a mutation that reduces but does not completely abolish function, consistent with the yeast and mammalian LCL data.
This is the first time a mutation in a nonredundant core protein involved in autophagy has been reported; however, other neurological disorders are affected by other proteins involved in the broader autophagy pathways. Autophagy is a process involved in all cells, but is needed in neurons more than in other cells. Although the patients with this mutation were classified as ataxic, and the main defects are in the cerebellum, pure ataxia does not involve intellectual disability, so, clearly, both the cerebellum and the brain in general are affected here. Autophagy has previously been implicated in neurodegenerative disorders such as Parkinson disease, and in spastic paraparesis, another movement disorder that has some overlapping symptoms with ataxia. Although we do not quite understand yet exactly why neurons are more vulnerable, there are many human disease mutations in ubiquitously expressed genes that nevertheless cause only a neuronal phenotype.
Autophagy has been implicated in ataxia before, primarily because the most common forms of ataxia, the dominant spinocerebellar ataxias, SCA1, SCA2, SCA3, SCA6, etc., are characterized by intracellular accumulation of aberrantly folded polyglutamine-containing proteins. These are thought to be toxic to the cell, evade autophagic sequestration and have even been postulated to inhibit autophagy. Our results demonstrate that decreased autophagy alone can cause ataxia even in the absence of abnormally expressed proteins, and hence suggest an alternative mechanism: The polyglutamine-containing proteins may be toxic to neurons by saturating the autophagy system, reducing the ability of the cell to deal with other critical processes (such as elimination of damaged mitochondria), rather than the aggregates being toxic per se. It has already been shown that increasing autophagy with drugs will improve symptoms in patients with ataxia having expanded polyglutamine-containing proteins. If reduction of autophagy becomes a more general theme of ataxias, therapies to increase autophagy may help ataxias more broadly, and our results would then have wider implications.
Funding
Funded by NIH grants NS0785600 (to MB), OD018265 (to JHL), GM053396 (to DJK), Wellcome Trust 087518/Z/08/Z and Lendulet LP2014-2 (GJ), Boğaziçi University Research Fund Grant 6655 (to AT) and HHMI and ALSAC/St. Jude (BAS).
Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed.