28 November 2011. The amyotrophic lateral sclerosis field is still searching for that super model. “There is no perfect mouse yet, but they all have interesting features,” said Amelie Gubitz, program director for research on ALS at the National Institute of Neurological Disorders and Stroke in Bethesda, Maryland. At RNA-Binding Proteins in Neurological Disease, a prelude to the Society for Neuroscience annual meeting held November 10-11 in Arlington, Virginia, attendees heard about four new mouse lines that produce various amounts of TAR DNA binding protein 43 (TDP-43).
Mutations in the RNA-binding protein TDP-43 cause amyotrophic lateral sclerosis (ALS) or frontotemporal lobar dementia (FTLD). In those diseases, TDP-43 exits the nucleus to form cytoplasmic aggregates. Philip Wong of Johns Hopkins University in Baltimore, Maryland, and Zuoshang Xu of the University of Massachusetts Medical School in Worcester each presented new model mice in which they attempted to mimic the loss of nuclear TDP-43 function by reducing or deleting the protein. Wong also described potential models for mild TDP-43 overexpression and for TDP-43 pathology in muscle tissue. And these four are not the only up-and-coming ALS mice out there: At the main Society meeting, held 12-16 November 2011 in Washington, DC, Peter Joyce of the Medical Research Council Mammalian Genetics Unit, Harwell, U.K., presented a new mouse based on superoxide dismutase 1, another aggregate-forming protein associated with ALS. The animals have a spontaneous mutation, in contrast to the usual SOD1 overexpressing strains.
At the Arlington meeting, chaired by Fen-Biao Gao of the University of Massachusetts Medical School in Worcester and Paul Taylor of St. Jude Children’s Research Hospital in Memphis, Tennessee, researchers noted that making ALS models by overexpressing TDP-43 might be the wrong way to go. “These overexpresser mice are pretty artificial,” Gubitz told ARF. Wong noted that most published TDP-43 mouse models make 2.5 to five times the normal amount of the protein (see ARF related news story on Wegorzewska et al., 2009; ARF news story on Wils et al., 2010; ARF news story on Xu et al., 2010; ARF related news story on Shan et al., 2010). Xu added, “We do not have credible evidence that TDP-43 [expression] is increased in humans.”
The problem in people, Xu and others believe, is nuclear depletion of TDP-43. One might assume, then, that a knockout model is appropriate—but Xu does not think there is absolutely no TDP-43 in the nucleus of ALS motor neurons. No null-TDP-43 mutations are associated with disease, he noted, and animals lacking any TDP-43 are not viable. He thinks instead there is less than normal TDP-43 in the nuclei of cells with ALS pathology. With this in mind, Xu set out to minimize, but not eliminate, TDP-43 expression using RNA interference. He introduced into mice a microRNA that interferes with TDP-43’s natural mRNA. By controlling the miRNA with the ubiquitous cytomegalovirus-chicken-β-actin promoter, Xu obtained a modest knockdown whereby TDP-43 mRNA dropped by half in the spinal cord and forebrain, and protein by some 20 percent in both the central nervous system and muscle tissue.
TDP-43 knockdown models are difficult to make, Xu told ARF, because TDP-43 regulates its own expression. Making a heterozygous mouse with one good and one bad copy of TDP-43 is ineffective because the protein simply upregulates its mRNA so that normal protein levels are still produced. The team’s RNA interference approach is a “creative” solution to this problem, commented Virginia Lee of the University of Pennsylvania in Philadelphia, who was not involved with the work.
Xu’s knockdowns started out normal—as do people who will go on to get ALS—but developed a wobbly gait around five to six weeks of age. This worsened to paralysis in some limbs by nine weeks. Internally, “there is exquisite selectivity to the motor neurons in terms of degeneration,” Xu said. Eventually, 60 percent of motor neurons were lost and the mice died around 100 days of age. “Because of the phenotypic resemblance to the human disease, this result suggests that a reduced function of TDP-43 may be the root of ALS and FTD,” Xu wrote in an e-mail to ARF. The model still needs further testing, he said.
Like Xu, researchers at the Wong lab worry that animals that highly overexpress TDP-43 make inaccurate models. These animals are so sick that they die too soon to properly mimic disease, said William Tsao, a student of Wong’s who presented a poster at the SfN conference. The animals do not even get paralysis. Even so, Tsao still thinks overexpression has some merit, saying that it needs to be done a little bit, instead of a lot. Tsao specifically selected transgenic mice that made slightly more of the protein than normal, using animals that carry either wild-type TDP-43 or the disease-linked mutant glycine-298-serine under the neural Thy1.2 promoter. His animals have approximately 1.5 times the normal amount of TDP-43 protein, including both the endogenous mouse version and the protein made from the human transgene. The mice did get sick, but Tsao had to be patient to see symptoms because they did not emerge until the mice were in their prime, Wong said. In this way, these animals model people with ALS, who do not develop symptoms until well into adulthood.
At one month, Tsao’s animals had abnormal reflexes in their hind limbs, and by one year of age they were not so much walking as paddling around their cages, with a swimming-like gait. By 18-20 months of age, many were paralyzed. The paralysis was frequently asymmetric, again akin to the human disease. At approximately two years old, the animals were so ill the researchers had to sacrifice them. The mice with mutant TDP-43 transgenes progressed in their disease a few months faster than those carrying wild-type TDP-43.
Looking postmortem, Tsao saw that neurons in these mice contained aggregates, but TDP-43 was not in them, leading Wong to conclude that TDP-43 inclusions may not be necessary for disease. He speculated that the aggregates may be made up of mitochondria, which have appeared in inclusions in other TDP-43 mice (see ARF related news story on Xu et al., 2010) . These animals offer a broader window in which to try treatments, Tsao said, and he thinks they might be better than other TDP-43 mice for testing drug treatments. Although the late-onset phenotypes are a practical challenge for researchers eager to obtain results, they may be more informative about the human disease, Gubitz said.
Wong also described his lab’s approach to selectively knock out TDP-43 to ask questions about the proteins’ role in particular tissues. The group uses the Cre-Lox system of gene deletion to knock out TDP-43 in specific cell types. “Does loss of TDP-43 in neurons contribute to disease?” Wong asked. Postdoctoral researcher Yun Ha Jeong aimed to answer this question in a poster at the Neuroscience meeting. She put Cre under control of the CaMKII promoter to ablate TDP-43 specifically in forebrain neurons. The researchers predicted that these mice should recapitulate the symptoms of frontotemporal lobar dementia caused by TDP-43 pathology in that part of the brain.
That prediction appears thus far to be right on. Although only two-thirds of the mice survived past their first month, those that did exhibited dramatic, age-dependent frontal atrophy, Wong told ARF. He is not sure why so many die young, but speculated that leaky Cre expression might knock out TDP-43 in more than forebrain neurons. Those that lived past their first month showed behavioral symptoms reminiscent of human FTLD. In mazes, they exhibited little anxiety or curiosity about their surroundings, unlike normal mice, Jeong said. She next plans to examine the social skills of the mutant animals, looking for a lack of interest in other mice. Apathy is a major FTLD symptom. “We have to be careful trying to assess [FTLD] in mouse models,” Wong noted. Nonetheless, he told ARF he thinks these mice are the best model for FTLD so far because others based on the genetic risk factors progranulin and tau have less frontal atrophy (Ghoshal et al., 2011; Kambe et al., 2011; Yin et al., 2010; Yin et al., 2010). Wong’s group is using the same approach to develop a line of mice lacking TDP-43 in motor neurons to mimic ALS.
In Arlington, Wong also asked if loss of TDP-43 in skeletal muscle could contribute to ALS. He presented a model designed to answer that query by expressing Cre under control of the myosin light chain promoter. These mice, made by graduate student Sophie Lin, consistently weighed less than controls, exhibited muscle degeneration, and died at four to five months of age. This suggests that some of the weakness in ALS could be due to problems in muscle rather than in motor neurons, Wong said. The TDP-43 knockout caused downregulation of Tbc1d1, which regulates glucose transporters on the cell surface (Chiang et al., 2010; Sakamoto and Holman, 2008). As a result, the mice have to obtain energy from fatty acids instead of glucose, limiting their caloric intake and weight gain. This metabolic phenotype may be relevant to human disease, where people with ALS have very little body fat, Gubitz commented (see ARF related news story).
Many labs are making TDP-43 mice, as well as mice that model pathology driven by FUS (see ARF related news story), another protein that is mutated in some ALS and FTLD cases. Together, they join the longtime favorite model in ALS research: mice that highly overexpress mutant human SOD1. However, those classic models have their own set of shortcomings; in particular, treatments that help SOD1 mice have not been successful in human trials (see ARF Webinar). The mice may have abnormalities related to the SOD1 overexpression that do not relate to human disease, said Joyce, who reported on a new breed of mSOD1 mice made by random mutagenesis. Among the 10,000 mutant mice archived at the MRC Mammalian Genetics Unit, one set had an aspartic acid-83-glycine mutation in SOD1. This corresponds to a human mutation (Millecamps et al., 2010) that interferes with the protein’s zinc binding, folding, and enzymatic activity (Krishnan et al., 2006).
Phenotypes for heterozygous mutant animals are mild. For homozygotes, “there seems to be a slowly progressive disease,” Joyce said. At 10-20 weeks of age, they developed hind limb tremors. Fifteen-week-old mice lost one-fifth of their spinal motor neurons. Shortly thereafter, Joyce observed the animals dropping their hips toward the ground. This behavior gradually worsened so that year-old animals—which were still alive, unlike many mSOD1 overexpression models—were jittery and did not move much about their enclosures. One-year-olds also showed gliosis. There were gender-specific phenotypes as well. Female mice were still alive at 18 months, but many of the males died of liver cancer starting around one year of age. This mimics a SOD1 knockout phenotype (Elchuri et al., 2005). Joyce next plans to look at the animals’ behavior and neuromuscular junctions.—Amber Dance.