Revealing the Inner Workings of Pain: A Conversation with Clifford Woolf

This is the second in a series of Forum interviews with PRF science advisors.

Clifford Woolf, MB, BCh, PhD focuses on the basic mechanisms of pain, and translating discoveries into new therapeutics and diagnostics. He is best known for the discovery of central sensitization, in which ongoing painful inputs render central neurons hyper-responsive and contribute to chronic pain. Woolf is now the director of the F. M. Kirby Neurobiology Center and the Program in Neurobiology at Children’s Hospital Boston, US. He is also a professor of neurology and neurobiology at Harvard Medical School, and a faculty member of the Harvard Stem Cell Institute. He also serves as an advisor to several biotechnology firms and drug companies. Megan Talkington spoke with Woolf about his lab’s current endeavors, his excitement about the prospects for new discoveries in pain, and the philosophy that drives his research. The following is an edited transcript of their conversation.

What are you focusing on in your research right now?

There are several things. One is the recognition that inflammatory pain, and almost certainly neuropathic pain, both result from an interaction between the immune and nervous systems. We’ve all known that, but we are starting to see a degree of convergence and interdependence that is greater than we’ve appreciated before. I’m very fortunate to have three research fellows in my lab who are immunologists, so we've been able to approach the study of pain more from an immunological perspective. There have been people like Jon Levine at the University of California, San Francisco, who have done that before, but it’s quite rare. I think most people using inflammatory models pay no attention to the actual inflammation—their whole emphasis is on changes in the nervous system, whereas I'm saying there is a lot to be learned about the actual inflammatory process.

There has been an emphasis on microglial involvement in neuropathic pain, to which we contributed, to some degree. Certainly that's interesting, but I've recently refocused my work more on what's happening in the periphery. We now have the tools to ask exactly which immune cells and mediators are involved during inflammation, how they change the properties of sensory neurons, and what is the temporal sequence of changes. We can see that there are different players at different times, and that inflammatory pain is not just pain in the presence of inflammation, but a very complex interaction between the two systems. From the basic science point of view, the complexity is very exciting. From a drug development point of view, the number of players makes it difficult. If you only hit one, there are many others still there.

On neuropathic pain, the most exciting thing that we are doing now is looking at temporal changes in expression of mRNA transcripts in the dorsal root ganglion and the dorsal horn during the evolution of pain. In mice after nerve injury, pain begins to appear at about three days, gets progressively worse, and then plateaus. Using ARRA [American Recovery and Reinvestment Act] stimulus funding, we’ve been able to do microarrays every eight hours from before the pain begins until it stabilizes at day 10. This has generated an amazing amount of data, as you can imagine, that we are in the middle of sorting out.

It's generating a movie of the change in the expression of the genome at the time of onset of neuropathic pain, and it’s going to give us a granularity of pain that, again, is going to reveal its enormous complexities. It's a bit of a relay race, passing the baton from one mechanism to another, from one cell type to another. We are going to have to look hard to unravel the key points where you can interfere and stop the process, but I think we will get a view of the pathways that was technically unimaginable a decade ago.

Your lab just published a study in which you used wheel running as a measure of pain in mice. Can you talk about that?

I think that Cobos et al., 2012 is one of the coolest studies we've done recently. It addresses the fundamental issue of how to measure pain in an animal. The standard way is to apply a stimulus and measure the response, which is usually reflexive. Many people, particularly Jeff Mogil and Frank Porreca, have recognized the limitation of this, and have tried to introduce operant methodologies to interrogate what the animal is feeling. But most of those methods are rather complex and difficult to perform in a routine high-throughput fashion.

So Enrique Cobos del Moral, a postdoc in my lab, took a different approach, using the fact that mice really love running in a wheel. The standard way of assessing activity is to leave a wheel in the animal’s cage and measure how much the animal runs in 24 hours, and it’s a nice metric. But Enrique found that if he only put the wheels in the cages for an hour, the mice get in there and run like crazy—they run about a kilometer in an hour. They love it.

Then we asked: If we produce mild inflammation in their hindquarters, does it affect their running? Sure enough, it causes a massive drop in activity, in their decision to run or not. But the duration is relatively short—about three days—whereas if you measure reflexive responses, the changes last seven days. So at day five or six, if you poke the animal with a von Frey hair, you’ll see that it has some discomfort. But on the same day, it is happy to run. The standard measures show something different from a choice that the animal makes about running, presumably a balance between its discomfort and the desire to run.

What makes it special is that, when we give anti-inflammatory drugs, we can block the reduction in running with very low doses of drug. In reflexive assays for analgesic activity, mice require doses of NSAIDs [non-steroidal anti-inflammatory drugs] or COX [cyclooxygenase] inhibitors that are 10 to 100 times higher than in humans, and the assumption has related that to the surface area and metabolic rate of mice. That turns out to be completely incorrect. We saw that we could block the reduction in running using the same doses, per kilogram, that work in humans. The studies that use reflexive measures require doses that no human will ever see—no person is going to be exposed to ibuprofen at levels that are required to affect the animal’s von Frey sensitivity. These are levels where you see toxicity and off-target effects.

For morphine, the same argument applies: The dose of morphine you need to obtain a change in reflexive responses in mice is typically about 5 milligrams per kilogram, whereas the standard dose of morphine in a patient is 10 milligrams total. So it is completely out of sync. In the wheel-running test, the mean effective dose is tiny—0.08 milligrams per kilogram. As you go up to 5 milligrams per kilogram, the animals can't run. They are out of it. They look like they’ve been out partying all night. In healthy animals, you can use this as an assay of CNS side effects. If the animal loves to run, and the drug is producing some CNS change, then they don’t run. It’s an amazing window into how the animal is feeling, by letting it make the decision to run or not.

The beauty of it is that this is a simple assay. You can run one mouse at a time, or you can run a hundred mice at a time. No one has to measure anything or make any judgments about the mouse’s behavior—the mouse does the choosing, and you just let the computer record when the mouse ran, and how fast it ran for how long.

We are pretty excited by this, because we think it could be a very simple way of measuring anti-inflammatory analgesic efficacy in a way that’s much more sensitive, much more specific, and much more clinically relevant than the standard measures.

What are the biggest issues you see plaguing the pain research field today?

I think that one of the big questions is whether our preclinical models are true surrogates of human disease. Many people are thinking about that, and it was the driver for our wheel-running study. Related to that is the lack of success in predicting whether preclinical efficacy is going to translate into analgesia in patients. This is something the field has to get to grips with: What do our models mean, and how can we use them more effectively?

I think one of the reasons the pharmaceutical industry has abandoned the field is because they think we haven’t dealt with the issue of animal models satisfactorily. That’s their commercial choice, but I think it is a disaster. They are getting out precisely at the time when we’re getting more sophisticated at asking what pain means at a neurobiological level and how to measure it, and I think, therefore, there will be success.

We are making progress in understanding pain and addressing the issues that will help us develop new therapies and new biomarkers. It’s going to be difficult, and there are going to be lots of disappointments, but I still think we're in a pretty good place, and it's going to get better.

Through all your studies on pain, have there been times when you were really surprised by what you’ve found?

Absolutely. Central sensitization was not something I went out looking for. My luck or skill was only that when something totally unexpected happened, I didn't discard it. It's when experiments work very well, and in the expected way, that you know you’re running behind the curve. That's technology as opposed to real science. Our hypotheses are generally so weak, so uninformed, that if everything conforms to them, it makes me suspicious.

From a philosophical point of view, I'm a real convert from doing only hypothesis-driven work, which is the standard approach to discovery. Genomewide screens or expression profiling, genomewide association studies (GWAS), and proteomics are difficult and complicated, but they break us out of the confines of just studying what we know by revealing the unexpected, what we don’t know.

Do you have any dream projects waiting to be done?

There are several—but I will restrict myself to two. First, there has not been a series of GWAS in pain, as has been done in other fields. The challenges are enormous in terms of phenotyping and the numbers of people that will be needed. But for a condition that has quite clearly been shown to have a very high heritability, it is a real shame that there has been no funding for this. But then again, maybe the time for GWAS has come and gone. With next-generation genome sequencing, the new technology has advanced greatly and can provide much more information that will let pain skip a generation and move directly to the next stage in the near future to identify “pain genes.”

The second one, which is more prospective, is a project my lab will be actively involved in. I think it should be possible to take fibroblasts from any human and dial up a neuron of choice—a nociceptor, or dorsal horn neuron, or interneuron, or projection neuron. Within the reasonably near future, I think, that is going to become the reality. So we will be able to study the nervous system using human neurons—to look at the excitability of human neurons, study neuronal proteins in their native environment, and compare neurons from a patient versus a control. That is going to be an amazing opportunity for modeling disease and screening for new therapies.

Is there a particular book or paper that you think everyone in the pain field ought to read?

I'm not a great sentimentalist. I think the history of science is interesting, but I can't think of anything that, in itself, has represented the field—the field is so wide.

But when I think about my own career, when I started out, the idea that we would find ion channels like TRPV1 and sodium channels like Nav1.7 that had specific and definable functions in pain seemed absolutely impossible. The notion was that pain was almost a philosophical state of mind. But we have now identified the channels that contribute to heat pain, and in which a mutation produces pain in individual patients. So if you were to force me to name one paper, I would say David Julius’s discovery of TRPV1 [Caterina et al., 1997], because it represented the transition from pain as a soft science to pain as a hard science. That discovery said that you can study pain at a biophysical level, you can identify the proteins, you can see what happens when they interact. That's just ordinary hard science. It’s not psychology, but I'm not denying for a moment that pain is also psychology. Meaning that pain is experienced by a patient, and it is colored by many other things. But there’s also a very neurobiological component to pain, and I think we can take apart the system and understand how it functions. That’s my mission.

In the debate about whether depressed individuals are more prone to get chronic pain, or whether chronic pain makes people depressed, I use the analogy of tuberculosis, or as it was known in the past, consumption. Before Mycobacterium tuberculosis was identified as the cause, people who got TB were considered aesthetes who needed to go to sanatoriums high in the Alps. But all of that disappeared once there was an antibiotic that killed the bacteria.

I think the same is going to be true for the vast majority of pain. I think we have the capacity to understand how chronic pain is generated, and that will lead us to targeted treatments. When we do that, many secondary problems like catastrophizing and depression and lack of sleep will likely fade away, because in most cases they follow the pain rather than drive it. And I think we’re getting there.

Thank you for speaking with me, and for sharing your optimism with our readers.

Thank you.

_________________________________________________

PRF Related Content:

News: Progress in Pain Genetics: Translational Value (19 Mar 2012)