20 September 2012. Mouse modelers wasted no time when, five years ago, the TAR DNA binding protein 43 proved not only to be mutated in some cases of amyotrophic lateral sclerosis (ALS), but also to be deposited broadly in inclusions in the majority of sporadic ALS, and even a subset of frontotemporal dementia (FTD). Researchers immediately hoped this discovery would lead to a powerful new disease model that could overcome the flaws of the tired workhorse of ALS preclinical studies. For more than a decade, ALS researchers had been limited to mice expressing mutant human superoxide dismutase 1 (SOD1), which represent but a small fraction of familial ALS cases. Worse, experiments in the SOD1 mice had failed to yield desperately needed medicines to slow the disease (see ARF Webinar). Many labs raced to engineer new mouse lines.
“We jumped into the TDP-43 modeling scenario … very quickly after the gene was discovered,” noted Jada Lewis, of the University of Florida in Gainesville, in a presentation at the 8th International Conference on Frontotemporal Dementias held 5-7 September 2012 in Manchester, U.K. But then—and now—researchers did not fully understand how TDP-43 contributes to ALS and FTD. Half a decade later, how are the new lines measuring up? Not quite as high as some had hoped. Scientists who are partway through characterizing the earliest models have discovered unexpected phenotypes that could nix their dream to test drugs in certain TDP-43 mice. But to be fair, Lewis noted that five years is a short time in the world of mouse modeling. In comparison, it took 20 years for scientists to come up with a good model for the hemoglobin mutations responsible for sickle cell anemia, she said. The take-home message for scientists is that rigorous characterization of a mouse model should precede its use in drug discovery.
With TDP-43, Robert Baloh of Cedars-Sinai Medical Center in Los Angeles was first out of the gates. His model expresses an alanine-315-threonine substitution also found in human ALS. Researchers greeted the publication with considerable excitement (see ARF related news story on Wegorzewska et al., 2009). “There were high expectations. People were thinking these mice would start getting flagrant FTD and ALS,” Baloh said. “The problem is, they are still mice. Humans and mice, even if they have the same genetic mutation, get different diseases.” In further analyses, researchers are finding that it is not neurodegeneration that kills Baloh’s mice. Instead, they die from bowel obstruction.
Disheartening revelations such as this are coming out as researchers breed the animals with standard strains and drill deeper into their pathology and symptoms than the paper initially reporting the new model was able to do. The Jackson Laboratory in Bar Harbor, Maine (JAX), breeds and distributes Baloh’s mice; they are also analyzing the animals’ phenotype.
The initial publication on a new model is just a description. At that stage, the mouse is not ready for drug studies, said Steven Perrin of the ALS Therapy Development Institute in Cambridge, Massachusetts. Like JAX, ALS TDI is breeding and characterizing the Baloh model. They are also doing that for another early model of the TDP-43 rush, from Jeffrey Elliott’s lab at the University of Texas Southwestern Medical Center in Dallas. Elliott’s group produced A315T, methionine-337-valine, and wild-type versions of TDP-43 mice; ALS TDI has started working with the wild-type so far (Stallings et al., 2010). ALS TDI expects to issue recommendations on how researchers should conduct preclinical studies with TDP-43 mice.
To date, at least 10 TDP-43 lines have been reported in the literature (reviewed in Cohen et al., 2011 and Wegorzewska and Baloh, 2011). As a group, they were difficult to obtain because the protein’s role in development makes knockouts difficult, and their ability to regulate their own protein level complicates overexpression. Broadly speaking, the lines share many features. Most seem akin to ALS, although researchers can push them toward an FTD phenotype by using a forebrain-specific promoter. The animals tend to develop trouble moving and die young. They suffer neurodegeneration and neuroinflammation. Ubiquitinated inclusions usually speckle their neurons. TDP-43 protein vacates the nucleus and fragments, though oddly, it does not join ubiquitin in those aggregates as frequently as it does in human tissue samples.
“The mice reproduce a great deal of the pathology … and a large part of the disease phenotype,” Elliott said. He thinks TDP-43 transgenic mice more closely resemble sporadic ALS than does the SOD1 model, because the mice get sick regardless of whether they express mutant or wild-type TDP-43. “That is an important feature,” Elliott said, since the majority of people with ALS have wild-type TDP-43. In contrast, mice with wild-type human SOD1 develop only minimal disease and live a full lifespan.
But reproducing pathology is only the first step. Subsequent treatment studies with a given mouse model also need to meet certain quality standards. For example, in the first decade of SOD1 mouse research, scientists across the field tested treatments with as few as four mice per treatment arm, Perrin told Alzforum. These studies generated papers in top-tier journals but did not reproduce in larger validation studies, much less in clinical trials. It was not until 2008, when ALS TDI published a study based on 5,429 mice from the SOD1-glycine-93-alanine strain, that researchers learned it would take at least 24 animals per group to robustly measure a 5 percent improvement in survival (Scott et al., 2008; see also ARF Webinar). Those groups should be balanced for males and females, because males sicken more quickly. Perrin further recommends leaving out animals that die early of something other than neurodegeneration, as that confounds results.
ALS TDI’s paper was “a big eye opener,” said Cat Lutz of JAX. Since 2008, awareness of a problem with reproducibility of mouse treatment studies has spread to the Alzheimer’s research community and beyond (see, e.g. (Shineman et al., 2011). Modern ALS studies tend to follow ALS TDI’s guidelines, but it is too soon to tell if the improvements in preclinical research herald success for clinical trials, Perrin said.
In applying the same rigor of characterization to TDP-43 mice, scientists at JAX and ALS TDI first had to stabilize the genetic background of Baloh’s animals. Scientists frequently generate new mutants in hybrid strains because these are convenient to work with, Lutz said. For example, their embryos have large nuclei that are easy to inject. But genetic background can greatly affect phenotype, so the ALS TDI and JAX teams bred the new model with standard black 6 (C57BL/6J) mice for several generations. Making Baloh’s mice congenic has made them better candidates for drug studies, Lutz said. Originally, their lifespan varied greatly, but moving the mutation into the black 6 background made it tighter and reproducible.
The first challenge in readying the line for drug research turned out to be simply breeding a sufficient number of mice. “That took a lot longer than we had anticipated,” said Fernando Vieira of ALS TDI. Typically, 80 percent of male-female pairs will yield litters, but for the TDP-43 mice, only 40 percent did. Part of the problem is that the transgenic males fell ill so quickly, they were not around very long to breed. At this point in ALS TDI’s studies, males appear to live for about 100 days; females typically die at 130 days but some make it out to 210, Vieira said. Wild-type lab mice live at least two years.
The researchers are unsure why breeding was so difficult. Vieira noted that the prion promoter controlling the transgene would turn on mutant TDP-43 in the testes; many males suffered odd abscesses near their genitals that may have killed the mood, so to speak. After struggling to produce approximately 300 mice the natural way, ALS TDI has resorted to in-vitro fertilization, and is in the process of obtaining 600 more animals, Vieira said.
Gut, Not Brain, Dooms ALS Mouse
To their surprise, the ALS TDI and JAX teams observed that the Baloh mice do not die of progressive neurodegeneration. Technicians were puzzled when they noticed animals went from health to death within a couple of days. The cause turned out to be bowel obstruction. “They feel lumpy,” Vieira said. “Their guts are completely filled with food.” Others have observed this as well (Guo et al., 2012).
The symptom is intestinal, but the root cause is likely neuronal. At JAX, researchers observed that the gut of the model mice might contain fewer autonomic ganglion cells, though Lutz cautioned that the finding is preliminary. This would suggest that the autonomic nervous system, which controls how the intestine moves food, stalls in the TDP-43 mice. “There is nothing wrong with the gut itself,” Baloh said. “It is a control problem of the gut.” This is not known to be a symptom of human ALS, although occasionally people do report digestive problems.
ALS TDI has not been studying Elliott's wild-type TDP-43 model as long as Baloh’s. Elliott’s also uses the prion promoter, but thus far the researchers have not seen bowel obstruction. “Why is the Elliott mouse different from the Baloh mouse?” Perrin asked. After all, the transgenic approach behind both is similar. He thinks it might have to do with the level of transgene expression—it is higher in Baloh’s strain—or with the locus where human TDP-43 settled into the mouse chromosome.
Lewis was not surprised to hear of the bowel phenotype. She noticed the same problem with mice expressing high concentrations of transgenic tau, a line that also relied on the prion promoter. Lewis suspects that something about neurodegenerative proteins under the control of this promoter attacks the neurons that regulate the gut.
Degeneration Versus Development
Because of the bowel blockage, Baloh’s mice never reach the point where they would suffer the full-blown muscle atrophy and neurodegeneration typical of ALS, Vieira said. In the spinal cord, it appears there are fewer motor neurons than normal, but those that are present show no signs of degeneration or apoptosis. He speculated the missing motor neurons might never have developed in the first place. “There is something going on there, but I do not know if it is actually motor neuron degeneration,” he said.
Like Baloh’s, most models are based on a TDP-43 transgene that is active from birth. These always-on transgenics can certainly tell scientists about TDP-43’s physiological role, Lewis said. “But for disease relevance, we need to get past what TDP-43 does in development.” Like so much about the protein’s function, it is not certain what it is doing in development, but the gene clearly is important since knockout embryos do not survive.
At the Alzheimer’s Association International Conference, held 14-19 July 2012 in Vancouver, Canada, Lewis presented an alternate transgenic approach from her paper in Acta Neuropathologica in June (Cannon et al., 2012). The team created a repressible TDP-43 model with the protein expressed in the forebrain, to mimic FTD, unless the animals are fed doxycycline.
Mice on a doxy-free diet looked much like other TDP-43 models. They had smaller brains, though the tissue did not shrink with age as would be expected for neurodegeneration. Instead, the growth rate of the transgenic mouse brains dropped between 12 and 24 days of age, when the brain is still developing. Other TDP-43 models also show low brain weight early in life (see ARF related news story on Xu et al., 2010; Xu et al., 2011).
When the team fed doxycycline to keep TDP-43 off until the mice were weaned at three weeks of age, they saw something quite different. The animals showed no physical symptoms, but in the brain, they progressively lost neurons, Lewis told Alzforum. Their brains atrophied to below normal weight by five and a half months of age, and continued to decline when the researchers checked at 11 months. In short, the pathology started to look like a neurodegenerative disease of aging such as FTD. The mice even showed inclusions positive for TDP-43 and ubiquitin. At the Vancouver meeting, Lewis encouraged other researchers with TDP-43 models to try the post-development variation.
TDP-43 Mice—Are They Good for Anything?
Clearly, there are plenty of differences between TDP-43 mice and human TDP-43 proteinopathies. “That does not mean the mice are useless,” Baloh said. “They can tell us pathways.” Elliott added that once researchers have a drug to test, it is worth trying it in a TDP-43 model. Given how many labs have generated and published TDP-43 mice by now, Elliott does not anticipate a much better model coming along.
Lutz advised that scientists fully understand the strains they are working with, and carefully choose questions the mice can accurately answer. For example, with Baloh’s model, the gut problem makes survival a poor outcome to measure. Yet researchers interested in other elements of disease, such as degeneration of neuromuscular junctions or how quickly the animals start dragging their limbs, might get just what they need out of this model.
To obtain robust results from Baloh’s model, Vieira cautioned scientists to stay away from the females. They live too long and their disease is too variable. Males could be useful—though not necessarily for ALS, he suggested; they might better model intestinal neuropathy. For his part, Vieira is still searching for the best TDP-43 model of ALS.
Whatever their worth as preclinical models, the TDP-43 and SOD1 mice are due to receive company. Researchers are working on animals transgenic for more recently discovered ALS genes such as fused in sarcoma and C9ORF72.—Amber Dance.