I'll describe this problem in the context of Huntington's disease (HD). HD is a truly devastating neurodegenerative disease that causes movement, psychological, and dementia problems.

What goes wrong in HD? In one way, we know exactly what goes wrong. HD is caused by a mutation in the gene that encodes a common protein called huntingtin (htt). However, there are many unknown aspects of how htt causes the disease.

Since htt is a protein, let's start by reviewing proteins. Proteins are the tools that the cell uses to actually do things. Proteins are made of little building blocks, called amino acids. When the amino acids are put together, they form a linear, one-dimensional sequence that lacks any function. However, each amino acid has a special affinity for other parts of the sequence, and so the protein folds into a complex, three-dimensional shape. Once this is accomplished, the protein immediately becomes functional. In this analogy, the protein's shape might program it to be a molecular hammer whose job it is to bang a nail.

To make our tool for this study, we used genetic engineering to attach a small green fluorescent protein to htt. Under a microscope, the entire neuron has a faint green tint because htt is found throughout the cell. The inclusion bodies can be easily seen as super bright green spots.

We used the labeled htt in conjunction with a robotic microscope that we developed. The microscope (pictured at left) can come back to exactly the same place on a plate of neurons multiple times over many days and can measure the same neuron over its entire lifetime. When the neuron dies, it disappears from the field, and we mark that as death. The microscope allows us to follow thousands of neurons and to measure when and if they formed an inclusion body and when they died.

From our analysis of hundreds of thousands of neurons, we were able to definitively say inclusion bodies are actually protective. This was a very surprising result, especially given the long-standing association of inclusion bodies with disease. But what was actually happening was that neurons that formed an inclusion body survived. Those that couldn't died. At autopsy, only the neurons with inclusion bodies were left alive and thus were incorrectly associated with disease.

So if inclusion bodies aren't the toxic species, what is? In that same study, we showed that the amount of diffuse htt, which was found throughout the cell was an excellent predictor of death. The more diffuse fraction Huntington you had, the faster the neuron died.

So we are still sorting the good guys from the bad guys in HD. Now we know inclusion bodies are good, and the bad guy or guys are in the diffuse fraction. As we noted earlier, mutant htt can assume a very large number of conformations. Which of those myriad forms is the bad guy?

Our task would be a lot easier if we could tag each htt form with a specific colored marker, but we don't have that technology yet. So we had to figure out a different way to develop a tool to measure the different forms in the diffuse fraction. And that tool was antibodies.

Antibodies are very useful. Of course, they are most useful to us to fight off foreign pathogens. If a virus or a bacterium infects us, we form antibodies, and they help to neutralize the pathogen. But antibodies are also extremely useful in biomedicine. There are millions of different types of antibodies, and each type recognizes and binds to a unique protein shape.

I'll use an example to illustrate this point. Unfortunately, the United States has a lower life expectancy than many European countries. We want to figure out why, but let's pretend we know nothing about modern medicine. We make several observations about the U.S. and Europe and try to associate them with life or death. For example, there's more obesity in the U.S. Fewer people smoke here, but they work more hours and have more options to buy shoes. Maybe these are risk factors.

In our analysis, if the U.S. has less of a factor, then it is probably beneficial, because we live a shorter amount of time. Conversely, factors that we have more of likely predict death. Without a better method, we might conclude that smoking is good. Just looking at risk factors and outcomes at the exact same time can create huge confounders to figure out which of these factors contributes, ameliorates, or is incidental to the disease.

What's the solution? Well, in clinical medicine, we use similar observational studies that involve a very large number of patients. At an early time point, we would measure their obesity, smoking, work hours, and whether they like shoes. At specific times, we would check them again and again, until they die.

These longitudinal studies allow us to pick apart the different risk factors and to determine which actually relate to death. By using a large population, we have lots of statistical power, and we can identify factors as detrimental, beneficial, or incidental.

How can we apply this same methodology to resolve problems in neurodegenerative diseases? How can we measure neurons growing in a culture dish in a laboratory?

This very question is a major issue in our laboratory. In essence, we want to do a “clinical trial” in a culture dish, and to do that, we need to follow the lives of thousands of neurons until they die.

The robotic microscope developed in the Finkbeiner lab has contributed significantly to the analysis of cell cultures.

To do this, we need several things. First, we need sophisticated tools that can detect and measure a specific risk factor. Next, we need to track the neuron over its entire lifetime. Finally, we need to be able to analyze all the complicated data to determine the relationships between certain risk factors and outcomes.

But in biology, nothing is simple. The problem with our strategy is that, for an antibody to bind its target, the tissue must be “fixed” in place by chemicals that kill the cell. You can detect what was in the neuron at the time that you add the antibodies, but then you can't follow the neurons anymore.

To get around this problem, we used some fancy tricks. We started by focusing on what we can measure longitudinally. We can tag htt with a green fluorescent marker, and we can follow the neurons over time. In fact, we have a choice of what type of htt we put into the neuron. We can put in a very toxic type of htt in one cell and measure when it dies. And we can put in a less toxic version of htt here and we know when it dies, and we put in the least toxic here and we know when it dies. Then we can add all three antibodies to a neuron and measure how much of each antibody binds.

The result is that we can determine how much of an htt type was in each neuron and when it died. By following two plates of neurons, we can produce a calibration curve. In one plate, we subject the neurons to survival analysis, our longitudinal method. In the other, we use our antibodies to determine how much of a specific form is present. By linking the two together, we can follow which forms of htt best predict death. And as it turns out, we've found one antibody that recognizes a shape that predicts death better than any other antibody.

In practical terms, with this information, we can learn more about that particular form. What other cellular components does it interact with? Where is it in the cell? Does it move? So even though we don't have the whole story, we're able to narrow down what the toxic species is and to focus our future studies appropriately.

At the same time, we can also do drug screens, in which a large number of potential drugs can be screened on the basis of how effectively they neutralize this particular form of htt. And in fact, we are working on exactly that analysis right now.

So, by being able to go from all of these potential suspects in the neuron and narrowing down to a very focused suspect, we’re able then to decipher not only why that suspect's toxic, but also start screening potential drugs against neurons and figuring out which of those drugs knocks down this suspect the most. And that has direct relevance for the disease and therapy.