This is the tenth in a series of Forum interviews with PRF science advisors.
Rohini Kuner, PhD, is a professor of pharmacology at the University of Heidelberg in Germany, where she studies signaling pathways that drive chronic pain. She has long focused on synaptic plasticity in the spinal cord, and her lab is also known for developing tools to create targeted changes in gene and protein expression in spinal cord circuits and primary sensory neurons in mice. Kuner recently overhauled her basic experimental approach: Instead of starting with mice and molecules, she now collaborates with clinicians to look first at patients and then explore mechanisms underlying clinically important pain conditions. Kuner spoke with Megan Talkington by phone about her new interest in pain processing in the brain, and about the responsibility of pain researchers to make sure their work is relevant to what happens in the clinic, among other topics. The following is an edited transcript of their conversation.
What is going on in your lab now that you are most excited about?
If I may, I would first like to give some history.
When I started my lab, I had, like many other people in the field, the dream of being able to work out how sensitization processes come about. The plan was to focus on the spinal cord, because it has a beautiful structure with fairly well worked-out anatomy and circuitry, and it’s possible to understand what is going on there at the synaptic level. What I really like is that it holds the first synapses, so to speak, where the peripheral nervous system meets the central nervous system, so it has to be a point where there is a lot of scope for modulation.
Our initial projects were directed towards working out the roles of receptors and signaling molecules that we took from the literature on the hippocampus, where we knew that synaptic plasticity occurs in learning and memory. In my postdoctoral work, I also had access to tools such as gene knockouts, so I could come up with hypotheses about spinal pain processing. That approach was very successful, and it led to very nice insights.
But I also realized that one would need a better handle for manipulating gene expression in the peripheral part of the pain pathway, namely, in the dorsal root ganglia; there were no tools for region-specific gene deletion. My lab generated a mouse that expressed the enzyme Cre recombinase specifically in peripheral nociceptors, which would enable us to work out how much pain processing was taking place at the peripheral level versus the spinal cord. At that time, the brain was an absolute black box for me—it’s something I found too complex to get into.
We used the Cre mice to address some candidate molecules we were interested in, such as cannabinoid receptors and glutamatergic receptors. I would define myself as a synapse person—a central nervous system person—so I was shocked at how much modulation occurs in the periphery.
One highlight was that, for analgesia evoked by cannabinoids—be they endocannabinoids or exogenously administered cannabinoids—we found, in collaboration with Clifford Woolf’s laboratory [at Children's Hospital Boston, US], that CB1 cannabinoid receptors expressed in peripheral neurons seemed to play a much bigger role than those in the central nervous system. This is consistent with new data from other labs studying cannabinoid-metabolizing enzymes such as FAAH [fatty acid amide hydrolase]. The picture that consistently comes out is that a large part of the analgesia from cannabinoids takes place at the peripheral level.
I’m very excited about this, because it enables the generation of new drugs—specific drugs that would affect cannabinoid signaling in the periphery. The advantage is that you would still evoke a pretty good level of analgesia without central side effects. That is one of the main problems with cannabinoids—the psychotropic effects and sedation. If you had asked most people a few years ago if they thought they could get around those problems, they would have said no, because cannabinoids work on the brain to produce analgesia, so of course one can’t dissociate the side effects from the analgesia. I think that this is indeed possible now. Most drug companies would throw away drugs that don’t enter the central nervous system, but in the context of cannabinoids, I would say that they should be looking for modulatory drugs that work in the periphery only.
What did you learn about glutamatergic receptors?
That work was along similar lines. I have a long history of studying glutamatergic signaling, both during my PhD work with Jerry Gebhart [at the University of Iowa, Iowa City, US, now at the University of Pittsburgh, Pennsylvania, US] and during my postdoctoral work with Peter Seeburg [at the Max Planck Institute for Medical Research, Heidelberg, Germany], who cloned most of the glutamatergic receptors. Everything was based on the premise that glutamatergic receptors play a key role in synaptic transmission, and that is very obvious both in basal transmission and plasticity. But I was always irritated because there were quite a number of studies from the 1970s showing that those receptors were also expressed in peripheral nerves. Mostly it was thought to be an artifact, because people thought that there is no synapse there, so where would the ligand come from? Where would you get glutamate?
We used our gene deletion approach to delete key glutamatergic receptor subunits in peripheral nerves. We found that a large part of the sensitization evoked by algogens, meaning sensitizing agents in inflammatory pain, results from an increased drive from AMPA glutamate receptors in peripheral nerves—so presumably there is glutamate that’s released from those nerves.
In collaboration with Gary Lewin’s lab [at the Max Delbrück Center for Molecular Medicine, Berlin, Germany], we worked out that when you activate TRPV1 there is a nice release of glutamate from the nerves, and this glutamate then acts locally, in an autocrine manner, on its own AMPA receptors in peripheral nerve endings. We think that excitation produced by TRPV1 is only partly from TRPV1—a big part of it comes from AMPA receptors that are activated by glutamate released in the periphery. We also found that if we blocked peripheral AMPA receptors, or deleted them by genetic methods, we could block more than half of the capsaicin response. That was very surprising to us.
Again, the lesson is that peripheral glutamate receptors determine how much drive comes from the periphery into the central nervous system, and they could be very good drug targets. AMPA receptor antagonists, especially calcium-permeable ones, are being developed by several drug companies for epilepsy and fear disorders. There you would need AMPA receptor antagonists that work in the central nervous system, so usually the companies would throw out drugs that do not enter it. But now we would say that, actually, one should keep those and study their potential as therapeutics in pain modulation.
The question of where glutamate comes from has been addressed recently by other groups who have shown that there is a lot of glutamate released from immune cells in inflammatory disorders, and high glutamate levels have been found in patients—for example, in the synovial fluid of the kneecap in arthritic patients. This all ties in beautifully with a role for peripheral glutamate.
That there is a lot of pain modulation at the level of the first gate into the nervous system—this has been an eye opener for me. If we can clamp down on signals in the periphery, very likely we can avoid long-term pain syndromes that involve changes in the central nervous system.
Have there been other developments that changed your thinking?
As a molecular pharmacologist, I have always looked at signaling pathways, and then asked whether a particular signaling pathway makes sense in a chronic pain paradigm. I had this tunnel vision of studying a molecule and then taking it all the way up to the systems level. We took up key molecules, perturbed them via molecular genetic methods, and then asked, How does neuronal activity change? How does synaptic transmission change? How does activity of the network change? And, finally, how does behavior change? From that we tried to conclude whether the molecules are interesting targets for therapeutic development. But recently, I’ve learned to think first about what really happens in patients, and then try to design pain models and experiments accordingly. It’s a reverse translation approach.
What inspired that shift?
I never had any insights into disease and pathology myself, but I got in touch with some excellent clinicians here in Heidelberg at the European Pancreas Center who are interested in the pathophysiology of pancreatic cancer and one of the major problems associated with it, which is excruciating pain. I didn’t have any molecular target in mind, but I really wanted to understand the disease and how pain could come about.
We looked at human biopsy samples and could see that patients with pancreatic carcinoma who also have a high degree of pain have major structural changes in their nerves: They have hypertrophic nerves, and a lot of nerve sprouting in the pancreas. We tried to figure out how this could come about. What could the candidate molecules be? We found that in almost all of the cases where there was a lot of pain, there was release of two cytokines, granulocyte colony-stimulating factor and granulocyte macrophage colony-stimulating factor—G-CSF and GM-CSF.
Those cytokines are hematopoietic growth factors, and they are depleted when patients undergo radiation treatment. G-CSF and GM-CSF are given to cancer patients when they are having radiation therapy because the patients have severely reduced blood counts and you need to increase hematopoiesis. But the main side effect of the therapy is pain. That really set us thinking. We had two links: Cancer patients with high pain levels showed high levels of G-CSF and GM-CSF in the pancreas, and giving those cytokines led to pain.
We asked whether it is possible that the receptors for those hematopoietic growth factors were expressed on sensory nerves, and they were—that was a big surprise. At first I was very concerned that it was an artifact, but we repeated it in several models, both in humans and in mice, and we saw reliable expression of the receptors on sensory nerves. We even saw that a signaling cascade downstream of G-CSF and GM-CSF, the JAK-STAT pathway, was activated in peripheral nerves and led to sensitization of nerves.
Then we asked whether blocking those receptors in the periphery could decrease cancer-associated pain. This was all very new to me, so I had a very smart medical student who went to Tony Dickenson’s lab [at University College London, UK] and Don Simone’s lab [at the University of Minnesota, Minneapolis, US] and learned how to use mouse cancer pain models. In those models, when we blocked G-CSF and GM-CSF signaling in the periphery, we could indeed considerably lessen cancer pain. But one could argue that the decrease in pain was because of a change in some immune component or in the tumor cells themselves, and so, as a crowning experiment, we downregulated the receptors specifically in nerves, and we again saw a major decrease in cancer-associated pain. This was a very interesting breakthrough, and I’d be so happy if companies develop it to treat cancer pain.
Cancer pain is now a very important issue for my lab. It is a very big problem, but there are not many people who work on it. Most people use models of inflammatory pain, because it’s easy to do in the lab, but the problem is that there is quite a lot out there already that works for inflammatory pain. It’s in disorders like cancer and neuropathies where more intensive research is required. One has to bite the bullet and take up complex models.
We are working now on models of pancreatic ductal adenocarcinoma (PDAC), diabetic neuropathic pain, and pain associated with spinal injury and demyelination. These are very complex disorders, but very important from the point of view of treating pain. Getting down to the pathological mechanisms in these disorders is very difficult, so it’s not the most rewarding work, but I think it’s something that has to be done. We work closely with clinicians, including oncologists, diabetologists, paraplegiologists, and psychiatrists, and that’s where the exciting part is—at the interface between neuroscience and other disciplines.
What is the key challenge? Developing the animal models?
Yes, and understanding the pathophysiology in those models. For example, there are forms of pancreatic cancer that are not painful. We’d like to figure out why only PDAC is painful; what is the difference between painful and non-painful types of cancer? When I tell people that we found this or that mediator to play a role in cancer pain, I often get asked if the mediator comes up only in cases of painful cancers, and I have to admit that I don’t have a single plausible answer yet.
You mentioned earlier that the brain used to look like a black box. How do you view it now?
This is a big challenge—the final frontier. There has been a lot of very beautiful work on descending modulation of pain—how the brain modulates pain processing in the spinal cord. But how pain is processed in the cortex has been studied very, very sparsely. Of course, cortical processing of pain, and the pain matrix, have been studied in humans with fMRI [functional magnetic resonance imaging] and PET [positron emission tomography], but not in animals, and in both cases, not at the level of single cells. For example, fMRI can tell you whether a particular brain region lights up in pain states, but it doesn’t tell you exactly which cells, how they are wired, or how their activity changes. How is pain encoded in a cortical network? What is it that, when it gets activated, says: I feel pain? Is it a subset of neurons that are pain-specific, or is it a combinatorial code of cells that get activated in a particular spatiotemporal pattern?
I think this processing needs to be understood because that is the key to central pain disorders. There are many pain disorders where there is no peripheral pathology, or the peripheral injury has long since healed but the brain has formed some imprint of pain—and this imprint, I feel, must be in the cortex.
What kind of experiments will help sort this out?
Excellent methods for studying cortical processing have been developed in the visual field—for example, two-photon imaging of structures and activities, such as with genetically encoded calcium indicators—but those tools have not been used extensively to study pain so far. This is something that my lab is now looking into.
We want to ask: How does the brain know that a person is sensing pain? Secondly, how does the cortical map change when there is a peripheral nerve lesion or a chronic muscle injury where there are aberrant pain sensations? One needs to look from both a structural and a functional point of view. Everything could be happening at the level of changes in synaptic spine structure, for example, or specific connections might change or re-form.
Studying this requires establishing really difficult methods in the mouse in the context of pain. It’s hard to get this kind of work funded, because it’s very risky, and it’s very fundamental. The European Research Council [ERC] has set up a program, called ERC Advanced Grants, which is dedicated to funding this kind of high-risk, high-gain research, and last year I was very lucky to get one of those grants. I now have several really excellent team members who are very motivated and are trying to establish these difficult methodologies to get this going. I don’t know whether we are going to be successful, but, if I could, I would love to address the question of how pain is encoded in the cortex.
Are there places where you feel the pain field is getting stuck and needs to change course?
I just returned from the European Pain School [in Siena, Italy], where we talked with young scholars about these issues. Again and again, we came back to the same conundrum: We don’t know whether what we are studying in animals, in terms of either pain models or pain behavior, really corresponds to what humans feel.
In animals, we can measure sensory deficits and sensory changes, but a large part of what patients feel probably comes from emotional modulation of pain—that’s one problem. A second problem is that most patients go to a doctor not because they have hyperalgesia, but because they have spontaneous pain—they have pain where they shouldn’t. This is very hard to study in animals, and it might explain why the scientific community keeps coming up with new mediators and signaling pathways and receptors that, based on our animal experiments, are supposed to play a role in pain, but then it’s very hard to translate that into the clinic.
I don’t know how to get around that issue, but we have to stop and analyze it. We can go on looking at new signaling pathways and pathophysiological mechanisms, but if they are not relevant to what happens in humans, I don’t know whether it is worth it.
How can pain researchers attack the problem?
The small change we made here was to say that we are not going to take, for example, a simple model of inflammatory pain and just crank through the next 50 knockouts that we have. If we did, we could have a nice long list of mediators and publish them, but it’s not very satisfying, because I don’t think that model is very close to the human condition. Instead, we want to take up complex pain states and try to make a model that is closer to what we see in the clinic. That takes quite a lot of work, but if I find something in that model, I will have much more confidence that it is relevant.
Other labs are trying to work out ways of measuring spontaneous pain, and I think that’s excellent; each person has his or her favorite approach. But I think each of us has almost a moral obligation to stop and ask whether what we’re doing is really relevant, and if not, how we can make it more relevant.
This is also what got us started working on the cortex. I find it fascinating that if you lesion a peripheral nerve, or if you have a spinal cord lesion after a car crash, or a lesion in the thalamus from a stroke—all of those lesions, at completely different levels of the pain pathway, induce very similar clinical symptoms. If the pathophysiological change driving neuropathic pain was mostly at the level of the peripheral terminal, or at the level of spinal circuitry, then I would not expect to see the same set of symptoms no matter where the lesion is. The common denominator in all of those neuropathic conditions is that inputs into the cortex have changed—maybe the cortex is trying to compensate for the lack of input but ends up doing so in an abnormal manner, or there is abnormal connectivity.
If we want to make real advances, then we have to look at what’s happening in patients and try to implement that in the questions we ask in animal models.
How can new pain researchers get started on the right track? Is there anything that you recommend they read?
I really admire the Textbook of Pain by Steve McMahon, Martin Koltzenburg, and others. We follow it like the Bible. For all new people who come to the lab, this is the book I put in their hands and tell them to take two weeks, don’t do anything else, and just read—it will be worth it.
I also tell my graduate students, who are often worried about competition from contemporaries and getting scooped, that the chance of that happening is small. The greater likelihood is that the question you are trying to address has already been addressed in the 1970s or 1980s. This happens! Obviously not with the same tools or molecules, because the field has had major advances in cellular and molecular biology. But in terms of pathophysiology, you come up with a new concept and think it’s your new concept, and you look at the work of your contemporaries and think that nobody is doing it. Then, just when you are deep into experiments, you realize, my God, in the 1970s or 1980s, a brilliant person who had very few tools at his or her disposal already had the idea.
I always ask my students—in whatever field they are working, in whatever model they are looking—to go back and look up studies from earlier times. I admire people like Pat Wall, Ed Perl, and others, who came up with so many startling concepts.
What else will help younger investigators?
I would like to praise the European Pain School in Siena that I mentioned. It’s a summer school organized by Manfred Zimmermann, Marshall Devor, Jordi Serra, and others, and it’s an excellent place where my students go once during their PhD. Every year they come back really enlightened, because they get a lot of basic knowledge, and also have a lot of debates. Often there are very senior people there; John Loeser, one of the founders of the IASP [International Association for the Study of Pain], was there this time. When you talk with these people, it’s a very humbling experience, because every time you think that we have developed a concept, you then see there was a lot known earlier, but it is very rewarding and motivating as well.
Thank you so much for sharing your insights with us.
It’s been great talking to you.
PRF Related Content:
News: EFIC VII: Taking Aim at Cancer and Bone Pain (12 Oct 2011)
Luo C, Gangadharan V, Bali KK, Xie RG, Agarwal N, Kurejova M, Tappe-Theodor A, Tegeder I, Feil S, Lewin G, Polgar E, Todd AM, Schlossmann J, Hofmann F, Liu DL, Hu SJ, Feil R, Kuner T, Kuner R. Presynaptically localized cyclic GMP-dependent protein kinase 1 is a key determinant of spinal synaptic potentiation and pain hypersensitivity. PLoS Biol. 2012 Mar;10(3):e1001283.
Gangadharan V, Wang R, Ulzhöfer B, Luo C, Bardoni R, Bali KK, Agarwal N, Tegeder I, Hildebrandt U, Nagy GG, Todd AJ, Ghirri A, Häussler A, Sprengel R, Seeburg PH, Macdermott AB, Lewin GR, Kuner R. Peripheral calcium-permeable AMPA receptors regulate chronic inflammatory pain in mice. J Clin Invest. 2011 Apr;121(4):1608-23.
Kuner R. Central mechanisms of pathological pain. Nat Med. 2010 Nov; 16(11):1258-66.
Schweizerhof M, Stösser S, Kurejova M, Njoo C, Gangadharan V, Agarwal N, Schmelz M, Bali KK, Michalski CW, Brugger S, Dickenson A, Simone DA, Kuner R. Hematopoietic colony-stimulating factors mediate tumor-nerve interactions and bone cancer pain. Nat Med. 2009 Jul; 15(7):802-7.
Agarwal N, Pacher P, Tegeder I, Amaya F, Constantin CE, Brenner GJ, Rubino T, Michalski CW, Marsicano G, Monory K, Mackie K, Marian C, Batkai S, Parolaro D, Fischer MJ, Reeh P, Kunos G, Kress M, Lutz B, Woolf CJ, et al. Cannabinoids mediate analgesia largely via peripheral type 1 cannabinoid receptors in nociceptors. Nat Neurosci. 2007 Jul; 10(7):870-9.
Tappe A, Klugmann M, Luo C, Hirlinger D, Agarwal N, Benrath J, Ehrengruber MU, During MJ, Kuner R. Synaptic scaffolding protein Homer1a protects against chronic inflammatory pain. Nat Med. 2006 Jun;12(6):677-81.
Other Forum Interviews:
Pain and Its Control: A Conversation with Allan Basbaum (6 June 2012)