It was while he was a neurosurgeon in training that Nicholas Boulis first witnessed just how complicated disturbances to the central nervous system can be.
"As I was working on spinal cord injuries, I became increasingly interested in ALS because it seemed like the cells in question were well-defined, and it seemed like there were options for delivering genes and proteins to motor neurons specifically," says Boulis, now an assistant professor at Emory University. "The more I worked on ALS, the more moved and motivated I became by it because I began to meet the families of people who were dying with it. It's a particularly compelling disease."
Amyotrophic lateral sclerosis — ALS — is a progressive, fatal neurodegenerative disease marked by the death of motor neurons, resulting in muscle weakness and eventual atrophy throughout the body. From the onset of symptoms, most people with ALS die within three to five years, typically from respiratory failure. While around 10 percent of ALS cases are heritable, the remaining 90 percent arise sporadically, appearing in people who have no family history of the disease. While researchers have made strides toward understanding the genetic factors that underlie the inherited, familial form of the disease since the early 1990s, the ultimate cause of ALS remains unknown. And for each of the more than 5,500 patients diagnosed with ALS every year in the US, there is no cure.
Despite this, the pace of ALS research has hardly been sluggish. In the last few years, investigators from the basic to the clinical ends of the research spectrum have come at the neurodegenerative disease from all sides, working together to translate bench discoveries into clinical therapies. During the last 12 months alone, they've identified a slew of novel ALS-associated genetic variants, generated a variety of new in vitro and in vivo models, unearthed cellular evidence as to the pathogenesis and progression of the disease, and performed innovative experimental surgeries that will inform future clinical trials of neuroprotective and regenerative therapeutic strategies.
In 1991, a team led by Northwestern University's Teepu Siddique localized four markers on the long arm of chromosome 21 as genetic causes of familial ALS. Later, Siddique and his collaborators identified the first ALS gene, reporting in a 1993 Nature paper a linkage they observed between SOD1 — a gene on 21q22.1 that encodes a Cu/Zn-binding superoxide dismutase — and the inherited form of the disease. The researchers also reported their identification of 11 such missense mutations in SOD1 among 13 families affected by familial ALS.
SOD1 stood alone for some time. Through 1998, researchers identified three additional loci associated with the familial form — which mapped to chromosomes 2q33, 15q15-q21, and 9q34 — but no other variants associated with ALS. But in 2001, Siddique and his colleagues identified deletion mutations in ALS2 that are associated with the recessive form of juvenile ALS and juvenile-onset primary lateral sclerosis.
Since then, researchers have discovered several additional mutations that segregate with motor neuron disease phenotype and familial ALS, including TDP-43 and FUS — which both encode RNA-binding proteins — and UBQLN2, which encodes the ubiquitin-like protein ubiquilin 2.
A number of linkage analyses emerged in the last decade, but two separate studies from 2006 that associated a region on the short arm of chromosome 9 with familial ALS and frontotemporal dementia captured the National Institutes of Health's Bryan Traynor's interest. Back in 2006, Traynor and his colleagues had experienced limited success identifying candidate genes for ALS, he says.
Traynor, who now heads the Neuromuscular Diseases Research Group at the National Institute on Aging's genetics lab, says the two 2006 papers "told us, first and foremost, that this was a common cause of ALS and FTD." Second, they narrowed the genomic region of interest for both diseases down to a tractable 7 million base pairs — which, in the geographical context of the entire human genome, "is about the size of a city," he adds.
Meanwhile, Traynor and his colleagues initiated a genome-wide association study of 442 Finnish patients diagnosed with ALS and 521 healthy controls of similar ancestry. "Finland was an ideal population for such a study [because] it has the highest incidence of ALS in the world, [and] it's very genetically homogeneous," he says. During their analysis, Traynor and his colleagues were not surprised to find "a spike on chromosome 21 corresponding to the SOD1 gene," he says. But, what they had not expected to see "was this huge spike on chromosome 9," which, he adds, "remains the highest spike that has been seen in any ALS GWAS to date." That hit, which localized to 9p21, "narrowed the region down from 7 million base pairs down to 232 kb," Traynor says. "Basically, we went down from a city to a street."
That data elated the researchers, leading them initially to believe they had only to identify which of the region's three coding genes had generated the signal. "When I saw this result, I said 'We'll know the cause of this in two weeks, because it'll take us that long to sequence all of the coding regions in the genes,'" Traynor recalls.
But that wasn't the case. The researchers' initial capillary sequencing experiments yielded no results. Next, "we went back and Sanger sequenced across the entire region — introns and intragenic regions — [and] didn't find anything," Traynor says.
Eventually, the team looked to the National Institutes of Health's intramural next-generation sequencing capacity to demystify the 9p21 signal. First, the team ordered Agilent SureSelect baits customized to the region. "What Agilent does is they go into the human reference genome and they design baits based on what is known about the region of chromosome 9," Traynor says. "But if you've got something in there that you don't know of, they're not going to design baits against it and you're not going to pull it down." Unable to identify the source of the signal using this approach, Traynor and his colleagues realized they were searching for a novel target.
Next, the team sent cell lines to the German molecular cytogenetics firm Chrombios, which flow-sorted the samples and returned them to Traynor's team. That DNA that consisted just of chromosome 9, he says. From there, "we basically sequenced the heck out of it. We got really high coverage across chromosome 9 ... we're talking 300x to 400x," he adds.
Given all the approaches they had tried to that point, Traynor and his colleagues felt fairly certain they were looking for either an inversion or a repeat expansion. When they didn't find an inversion after filtering their data for novel variants, the researchers homed in on a repeat. "Within the core region — that 232-kb block of linkage disequilibrium — we saw eight novel variants. Of those, six were all within a 30-base-pair region," he says. Using the Broad Institute's Integrative Genomics Viewer, the researchers interrogated that 30-bp slice and were surprised to find a dramatic drop in sequencing coverage over it. "It basically went: 300x, 300x, 300x, then dropped down ... and came back up to 300x," Traynor says. "It all came down to two lousy sequences." Of the billions they'd generated during the course of their investigation, "two reads were really what gave it to us," he adds.
In September, Traynor's group and another team — led by Rosa Rademakers, an associate professor of molecular neuroscience at the Mayo Clinic in Jacksonville, Fla. — both reported in Neuron their independent identification of a hexanucleotide repeat expansion — GGGCC — in the non-coding region of C9ORF72 as the cause of 9p21-linked ALS-frontotemporal dementia.
"We now can explain 50 percent of familial ALS in Finland and about 40 percent of familial ALS elsewhere," Traynor says of his team's study. "Because of that, we've now gone from understanding a quarter of familial ALS ... to two-thirds in terms of genetics."
Similarly, Rademakers says her group's discovery "provides us with new clues about the disease mechanisms underlying FTD and ALS and provides new avenues in which to perform research." She says that it's "currently unclear whether C9ORF72 plays a role in the ALS-FTD disease process." To study the functional consequences of the gene, Rademakers adds, researchers will "likely start to develop mice carrying the human C9ORF72 transgene and C9ORF72 knock-out mice. We first need to understand more about the normal function of C9ORF72 before we can determine its role in disease."
Likewise, Traynor says that the National Institute on Aging's cell biology group and transgenic mouse core are both "very interested in ... making animal models and going after this" repeat expansion.
As translational research goes, animal models have been instrumental tools for many diseases. "Understanding disease mechanisms in model systems — mouse, rat, fly, and zebrafish — can be very important as tools," says Lucie Bruijn, chief scientist at the ALS Association.
For some neurodegenerative diseases, however, the utility of transgenic animal models has been contested. Still, according to Zuoshang Xu at the University of Massachusetts Medical School, the potential limitations of mouse models have made them no less important for moving ALS research along.
"For every new gene that's being found, people naturally start building transgenic models," Xu says. "That's one of the key reagents that we will need" going forward, he says, because "in vitro findings need to be validated in vivo."
Xu adds that in vivo models are critical to establish "whether a particular gene mutation is causing a gain-of-function type of toxicity, or a dominant inactive, or a loss of function, because each of these mechanisms will require different therapeutic strategies." He says that in vitro models, while informative, are not suitable for this purpose because they leave researchers to wonder whether "this causes this in this condition, or that causes toxicity in that condition, [or] a particular experimental condition causes the phenotype."
But animal models are not without their problems. Brian Kaspar, principal investigator of neuromuscular disorders in the Center for Gene Therapy at Nationwide Children's Hospital in Columbus, Ohio, says "most things that have been tested in the mouse have actually failed in human clinical trials." And since most manipulations have failed for no apparent reason, researchers are left to wonder: "Is it the mouse model, or is it the design of the study in the mouse?" Kaspar says.
"The over-expression approach has been really generating a lot of confusion," Xu says. That's because it's tough to tell whether a toxicity phenotype is purely the result of mutant over-expression or a function of the wild type in response to it, he adds. Further, while some ALS animal models do indeed show a neurodegenerative phenotype, Xu says a similar outcome might be achieved simply by expressing any toxic protein. With transgenic mouse models, overall, "the main difficulty with this is, in humans, nobody has reported an increased expression of TDP43," he adds. "There's no evidence that the over-expression condition exists in humans."
Kaspar says some animal models for ALS show an unrealistically exaggerated mutational over-expression. He says that one of the more commonly used SOD1 transgenic mouse models expresses "20-times more SOD1 than a human would."
Knockout mice can also be problematic. While a homozygous mutant may be lethal, its heterozygous counterparts might not display an affected phenotype. In general, "knockout mice have many abnormalities and they don't live long," Xu says.
"People have argued upon the importance and usefulness of the mouse because things have failed," Kaspar adds. Despite this, he says he still considers "mouse models an important aspect towards therapeutic testing in this disease."
For Xu, improved animal models — perhaps "partial loss-of-function" mutants, he says — must be met with improved in vitro models. "It's a complementary approach — you can't do everything in vivo, and likewise, you can't do everything in vitro. In vitro findings need to be validated in vivo, and in vivo findings probably need in vitro systems to explore the details of molecular mechanisms," he says. Overall, he adds, "the critical question is what is the mechanism by which these mutations cause neurodegeration?"
As such, the ALS Association's Bruijn says research findings from animal models "must be validated looking at human materials where possible."
(Cell) culture shift
Just as GWAS and sequencing have helped researchers unearth new genetic targets, ALS research has also seen a shift on the cellular level. Increasingly, investigators are turning their attention to cell types other than those affected by the disease for potential clues as to its pathogenesis.
"The field has really changed on the aspect that yes, it's a motor neuron disease ... however, recent work has demonstrated that there are other cell types that are as important, if not potentially the instigators for motor neurons to die, that really drive disease progression," Nationwide Children's Hospital's Kaspar says. Between astrocytes, microglial cells, and oligodendroglial precursors, "now we have new cellular targets," he adds.
In August, Kaspar and his colleagues reported in Nature Biotechnology that cultured astrocytes derived from both familial and sporadic ALS patient cadavers are toxic to motor neurons. "We asked the question: In a cell type that has been implicated in the rodent models, how did that translate? ... Was it really implicated in the human side of the disease?" Kaspar says. His team also sought to evaluate whether, "in the SOD1 cases where there's the normal level of SOD1 versus 20-times more expression, did one still see astrocyte-mediated toxicity?" he adds.
Indeed, in their paper, Kaspar and his colleagues show that SOD1 knockdown in vitro attenuates astrocyte-mediated toxicity toward motor neurons. Further, the team proposes its in vitro model system could facilitate future investigations into common disease mechanisms shared by familial and sporadic ALS as well as potential therapies for both.
"What's going on with these astrocytes? What are the molecular mechanisms at play here that aggravate them so much? What is it about motor neurons that make them susceptible to this toxicity?" Kaspar asks. His team now seeks to achieve "a greater understanding of the mechanistic aspects involved in these cells we've cultured," he adds.
In October, researchers at Johns Hopkins University School of Medicine reported their transplantation of SOD1G93A precursor cells into the cervical spinal cords of wild-type rats, in an effort to overcome the issue of ubiquitous over-expression of mutant SOD1 in most transgenic animal models studied. Writing in PNAS, the team said that the transplanted glial-restricted SOD1G93A precursor cells not only survived and differentiated into astrocytes, they also induced motor neuron death through an apparent host microglial activation. The Hopkins team said its study showed that mutant SOD1 astrocytes alone drive motor neuron death in vivo.
Cell, gene therapies
Transplanting potentially therapeutic progenitor cells is not a new concept. Indeed, Emory's Boulis and his colleagues have been performing clinical stem cell transplants since October 2010.
While the eventual goal is to replace damaged cells, "most of what we're focused on right now is protecting neurons," Boulis says. "We're using fetal-derived neural progenitors that come from the spinal cord — and which have been shown in rat models to protect motor neurons by a number of mechanisms — and transplanting those in so they're in close proximity [to the damaged cells]." He adds that directly injecting stem cells into the spinal cord seems to be "a very feasible option for ALS. Right now, we're pursuing that as our primary strategy for near-term translation."
The success of any cell therapy is limited to that of its delivery. "Direct injection paradigms are segmental, meaning you're only treating a limited segment or set of segments of the spinal cord," Boulis says. Still, he says a segmental approach is worth pursuing while more diffuse delivery approaches — which aim to treat all affected motor neurons rather than only those close to the injection site — are developed and tested because "there's less barrier to moving to the clinic with a direct injection paradigm right now. It's justified in the near term," he adds, "because ALS patients are dying from loss of diaphragmatic function and loss of control of their upper airways, so we can define a limited number of segments of the spinal cord — namely the cervical spinal cord — where treatment might prolong their survival."
For Boulis, that he and his team got the regulatory green light to "move into humans with stem cells with direct injection ... is making it easier to move ahead with the gene therapies."
Like many groups, the Boulis lab has devoted much of its gene therapy efforts toward retrograde axonal transport using viral vectors. "We have looked at different vector types, including herpes and lentivirus as well as … various serotypes of AAV [adeno-associated virus]," Boulis says. "Recently, we've begun to look at the use of AAV9 as a serotype and a couple of other serotypes of AAV with interthecal injections, and there are others who are pursuing intravascular injections," he adds.
Xu's group at UMass is also using AAVs as a therapeutic vehicle, though for RNAi delivery. While they have yet to demonstrate efficacy in human cells, Xu and his colleagues have shown that AAV serotypes 8, 9, and 10 "have the property of broad spread in the central nervous system" when delivered to SOD1 mutant mice. "The idea is to screen AAV serotypes that have the property of broad spread in the central nervous system," he says. That multiple serotypes have shown to be efficacious so far in mice is important "because you might imagine a situation where you give a patient a dose of therapy and for some reason it went half-way," he says. In that situation, "you cannot give the same serotype again because the patient would develop an immune reaction to it and later neutralize the second therapy. With multiple serotypes, potentially you can administer different serotypes to kind of bring it closer to the full therapeutic potential," he adds.
Still, sporadic ALS remains somewhat of a genetic black box. Emory's Boulis says a common flaw among most gene therapy studies is that "it's very attractive to pursue the familial forms of ALS because we know what genes to target. Problem is that's not most of the ALS patients." In addition, "all of our data on proof of principle is coming from a model of familial ALS, so it's unsettling that we're basing our ideas about what to do in sporadic ALS based on familial ALS," he adds. "They are phenotypically similar, but we don't know that they are similar, so that may be one reason for failure."
ALS research on the ascent
Despite the challenges, the National Institute on Aging's Traynor says the "pace of discovery is really accelerating. It used to go a year, two, or many years between genetic discoveries, and now it's shrunk down to a period of months between genetic discoveries." With funding from the Robert Packard Center for ALS Research at Johns Hopkins, Traynor and his colleagues aim to fully sequence 21 ALS patient exomes and make that data public.
For its part, Rademakers' group at the Mayo Clinic "will continue to characterize our large series of DNA samples of FTD and ALS patients for this mutation to provide more accurate mutation frequencies and to better define the clinical and pathological spectrum associated with [the] repeat expansion" it recently identified, she says. "We are also interested in determining the minimal repeat length needed to develop symptoms and, moreover, we are interested in identifying genetic modifying factors which could determine whether mutation carriers develop FTD, ALS or both."
According to Nationwide Children's Hospital's Kaspar, the next big challenge for geneticists will be to "really discern the genetic components involved in all the cases of ALS," he says. "That's a daunting task on the aspect that it looks like there are many genes that could be potential susceptibility genes, and all of these mutations that they're finding only account for a small fraction of the patients, but it's certainly taking us one step closer to really understanding the complexity, but also potential common molecular pathways that could be involved in the disease."
UMass' Xu expects high-throughput sequencing will continue to be an enabling technology for ALS -research. "Once all the single gene mutations are defined, sequencing will still play a very, very important role in defining other risk factors," he says. For example, sporadic cases of ALS "could be [the result of] multi-gene effects — potentially you could have two or three genes that have to combine to contribute to cause sporadic ALS." In addition, Xu expects transcriptomics studies using RNA-seq will be critical for defining changes in splicing, particularly in ALS-associated RNA-binding protein variants at FUS and TDP-43.
Emory's Boulis says that while "animal models give us reason to be enthusiastic and have created sufficient justification to move into clinical trials," he says they should "really be viewed as human experiments more than therapies that are ready to go. We've just done as much as we can do in animal models, and to move further is going to require human trials ... but nothing's ready for prime time yet."
When it comes to determining patients' tolerance of immunosuppression, or the proper methodologies for transplantations or injections, for example, "sometimes it takes doing something to optimize it," Boulis says.
Because of this, he adds, "the people who are volunteering for these trials are enormously heroic. I tend to view them as astronauts — they're doing something that is dangerous, but is necessary."