Where Are Glia Going?

When pain researchers ponder cells in the central nervous system that contribute to chronic pain, their thoughts generally turn to just one culprit: neurons. But studies over the last decade say that there are many more cells involved—in particular, the microglia and astrocytes—and that we ought not overlook them.

A growing body of data indicates that these cells, often loosely lumped together under the moniker “glia,” modulate neuronal excitability following damage and disease. The dizzying array of activities that microglia and astrocytes take on, and the complex signaling pathways involved, provide a wealth of potential entry points for therapeutics. However, most of the existing data come from rodent models, and in humans, early returns on one glia-targeted therapy were disappointing. Thus, while glia almost undoubtedly play major roles in pain in animal models, whether they are useful targets for human therapy is still an open question.

When neurons feel pain, their glial neighbors feel it too

Nerve injury evokes a response from a variety of immune cells in the periphery, which release inflammatory mediators that produce pain and sensitize peripheral nociceptors to further input. That is well known, but more recently, researchers have appreciated that something similar seems to be going on in the central nervous system. Microglia—the resident immune cells of the CNS—as well as astrocytes are now believed to be intimately involved in the synaptic plasticity that underlies central sensitization.

Linda Watkins, University of Colorado, Boulder, whose work has played a major role in revealing a role for CNS glial cells in pain, says that the first hints came from outside the pain field. Researchers found that astrocytes from the spinal cord of rats were responsive to substance P, a known pain neurotransmitter (Marriott et al., 1991). Meanwhile, Susan Carlton and her colleagues at the University of Texas Medical Branch in Galveston saw that damage to peripheral nerves in rats triggered astrocytes in the dorsal horn of the spinal cord to fatten up and multiply (Garrison et al., 1991). Watkins says that her own research connecting glia with pain was inspired by studies showing that physiological responses common to many illnesses—including fatigue, fever, and even pain—are driven by glial cells and their pro-inflammatory products. She and others found that central glia are involved in not only these physiological responses, but also pathological pain in rodent models. In particular, there is evidence that pro-inflammatory cytokines from dorsal horn glia help to produce pain hypersensitivity from inflammation and nerve injury (see Watkins and Maier, 2003).

Those and other converging lines of evidence eventually produced the widely accepted model in which peripheral nerve injury and inflammation send microglia and astrocytes in the spinal dorsal horn into an activated state. Those activated glia help to ramp up the excitability of neurons, contributing to chronic pain. The case solidified for a causal role of microglia proteins in pain hypersensitivity in 2003, when two labs showed that the MAP kinase p38 and the ATP receptor P2X4 are induced in spinal microglia after nerve injury and contribute to the development of allodynia in rats (Jin et al., 2003; Tsuda et al., 2003, respectively). Since then, glial researchers have gone on to describe complex signaling loops by which injury-activated microglia and astrocytes release trophic factors, inflammatory cytokines, chemokines, and proteases that lead to, and maintain, the sensitization of neurons. Many of these mediators are seen as promising targets for therapy, either through small molecules or gene therapy approaches (see Glorioso and Fink, 2009).

Notably, microglia activated by nerve injury or inflammation have different morphology, gene expression, and biochemical activities from the cells in their usual state. That means therapies aimed at bringing glia back to normal should ease pain while sparing other forms of sensation.

As the intricacies of glial biology continue to get worked out, a number of unexpected features have emerged that could be exploited for pain therapy. For one, it appears that glia are activated not only by neural damage and pain signaling, but by opioids, and this activation appears to lie at the heart of many of the limitations of opioids including drug tolerance and dependence, constipation, itch, and respiratory depression. Targeting glia therefore may offer a way to boost the efficacy of opioids. For example, opioid-induced glial activation is mediated at least in part by toll-like receptor 4 (TLR4), so Watkins and others are seeking inhibitors of that immune receptor as potential opioid adjuvants (Watkins et al., 2009).

With evidence like this piling up, it can be easy to think of glia as nothing but bad, at least when it comes to pain. But glia are clearly complicated characters: They ramp up neuronal excitability at some times, but put on the brakes at others. Ru-Rong Ji at Brigham and Women’s Hospital in Boston, Massachusetts and his colleagues have found that resolvins, a group of anti-inflammatory lipid mediators, reverse the plastic synaptic changes in the spinal cord that generate pain hypersensitivity (Xu et al., 2010). More recently, Ji says, he has found evidence that resolvins are made by central glia.

The first trial

Despite the many details left to be settled, glia-targeting agents are starting to move into clinical trials for pain. The first such trial, of propentofylline, was completed in 2009, and the compound failed to show efficacy. As a result, the fundamental strategy of targeting glia for pain treatment has been the subject of intense scrutiny and soul-searching among the research community.  

Propentofylline is a xanthine derivative that previously progressed to a phase IIIb trial for Alzheimer’s disease, but it was stopped in 2000 when patients on the drug failed to show improvement over placebo (for details, see the propentofylline entry on the Alzheimer Research Forum or Frampton et al., 2003). The drug’s precise mechanism of action is unclear, although it is known to be a phosphodiesterase inhibitor and to have neuroprotective and anti-inflammatory effects in vitro and in vivo. The specific cells it acts on, whether neurons, microglia, astrocytes, or others, are likewise not known, but propentofylline is commonly accepted to rein in glial activation. (For a review, see Sweitzer and De Leo, 2011.)

Joyce DeLeo at Dartmouth Medical School, Hanover, New Hampshire has also studied the effect of propentofylline in a variety of rodent pain models and produced preclinical data showing, for example, that propentofylline could reduce pain hypersensitivity from peripheral nerve injury in rats (Tawfik et al., 2008). Based on those results and others, in 2009 Solace Pharmaceuticals conducted a phase IIa study (ClinicalTrials.gov NCT00813826) to evaluate the efficacy of the drug (now called SLC022) in about 180 patients with postherpetic neuralgia. The results were disappointing: Propentofylline failed to decrease patients’ self-reported pain compared to placebo. Soon after, the company disbanded.

Compounding her disappointment at the trial’s failure, DeLeo says that after the breakup of Solace she has been unable to access the raw data from the trial, but says she is now working on a paper to report the top-line data she has. DeLeo presented the data on several occasions in the fall of 2010, which has given the field an opportunity to chew on the results and consider what they may mean for the future of glia and pain. The messages different researchers take from the trial could hardly be more divergent.

Some see the propentofylline story as nothing more than an isolated negative result. Watkins, for one, cites several caveats that might account for the trial’s failure. One concern is that the compound may have failed to reach the spinal cord in sufficient quantities and for sufficient duration. Another is that postherpetic neuralgia (PHN) may simply have been the wrong disease to use as the first test case for glia-targeted therapy. While glial responses appear to be central to a wide variety of pain conditions in animal models, PHN is one condition for which that has not been shown. Given those issues, Watkins told PRF, “I don’t think the trial failed.” Instead, she said, “It was a very unfortunate trial in a lot of ways,” and “It left a very big unanswered question” as to whether propentofylline and other glia-modulating compounds can succeed in alleviating pain.

DeLeo is far less hopeful. Looking back, she acknowledges that she would have liked to have tested the drug in patients with several different pain conditions, rather than focusing on a single condition. The choice, she says, was driven by the FDA’s desire to test treatments for particularly refractory types of neuropathic pain including PHN. Even so, she says, the failure of her group’s trial is making her question the fundamental premise that modulating glial activity is a good strategy for pain therapy.

Clifford Woolf at Children’s Hospital Boston, who with DeLeo served on the scientific advisory board of Solace Pharmaceuticals, is similarly grave about the trial’s implications. He acknowledges the possibility that the molecule did not make it to its targets in the spinal cord. However, outside of that sole caveat, he says, “The conclusion that has to be drawn is that microglial activation is not involved in chronic neuropathic pain.” In an email, Woolf added, “It is still possible, though, that microglia have a role in the acute induction phase of neuropathic pain, since that is the time when animal models are tested, but this will be very difficult to test in a clinical trial without knowing who is at risk of developing chronic neuropathic pain.”

DeLeo says glia researchers need to take the failure of the propentofylline trial to heart.

“I think we really should step back and think carefully about what we’re doing, and not generate the same type of data we’ve been generating—which is to use propentofylline or minocycline or any of these ‘glial modulators’ in an animal model, show a decrease in expression of cell markers, and then conclude that obviously this drug is going to have use in neuropathic pain. There are too many disconnects.”

“I think if people can chat about it, be more introspective… it would really help the field,” DeLeo says. “I’ve seen a lot of people saying, ‘Well, this glial thing was hot, but maybe it’s not as hot as we thought.’ And I think that’s good.”

Connecting the dots

The first gap that needs to get filled in, DeLeo says, is the space between rodents and humans. Almost all of the existing information on glia in pain biology comes from rodent models, so the first order of business, she says, is to determine whether rodent and human glial cells in culture respond to inflammatory stimuli in the same way. She says that her lab is now making those comparisons, and they are finding “vast differences.” Then, she says, more work needs to be done to determine how glia-modulating drugs affect human vs. rodent cells.

While DeLeo and Woolf think that the failure of their propentofylline trial is bad news for the prospects of glia-modulating drugs for pain, and Watkins still sees plenty of room to be hopeful, Ru-Rong Ji has a third perspective. The trial did nothing to dissuade him from believing that glia are intimately involved in chronic pain, and that altering their activity is a promising approach to treatment. But, he says, glial activities are so complex and multifaceted—they release both pro- and anti-inflammatory mediators, for example—that figuring out how to intervene is probably going to require more subtle approaches. Meddling with glia could easily backfire by wiping out those cells’ own pain-resolving functions. “We need a specific rather than a general inhibitor of the functions of glia,” Ji says. “We need more work to figure out the detailed mechanisms.”

Regardless of whether they are gloomy or giddy about the readiness of glia-pain biology to translate into therapy, glial researchers seem unified on one point: Their field has only seen the tip of the iceberg when it comes to basic biology. For starters, there is no definition of what “glial activation” really means. Viewing the cells as a binary system, in which glia are either on or off, is probably too simplistic. Different experiments score glial activation based on a wide array of markers and functions: One study may look at changes in cell morphology, while others involve calcium imaging or immunohistochemistry, and others follow patterns of gene expression or the release of pro- and anti-inflammatory mediators in cerebrospinal fluid. “‘Activation’ is a terrible term,” says Watkins, because it is a one-size-fits-all word used to describe changes observed by all of these different experimental methods. She says there is a great need for researchers from different methodological camps to draw connections between their results.

If “activation” is a fuzzy concept, “glia” may be even more so. Researchers frequently find it impossible to know whether the effects they observe are due to microglia, astrocytes, or even neurons. “We have no selective microglial inhibitor. This is a big problem,” says Watkins. What’s more, there may be other types of immunocompetent cells, beyond microglia and astrocytes, that ought to be considered. “Astrocytes and microglia happen to be extroverts under a microscope—when they become activated, they like to tell you about it by upregulating activation markers you can stain for,” says Watkins. “That doesn’t mean endothelial cells aren’t also activated.”

Making the leap into humans

As investigators scramble to develop a more detailed picture of glial biology, they frequently recognize that for the most part, they are painting that portrait in animal models, not in humans. So they are hungry to get some information on what happens in human beings. “The plain fact is that to date we have no direct evidence for a role of central glia in human chronic pain states,” wrote Stephen McMahon and Marzia Malcangio at King’s College London in the UK, in a 2009 review (McMahon and Malcangio, 2009). Today, Malcangio told PRF, the situation is no different.

Getting evidence of glial activation in humans has proven to be a tall order. One possibility is to analyze postmortem tissue. Indeed, one recent study used immunohistochemistry techniques to analyze autopsy tissue from the spinal cord of a patient with complex regional pain syndrome (CRPS) and found evidence of microglia and astrocyte activation (Del Valle et al., 2009). However, postmortem studies face the challenge of numbers. Often the study involves only a single patient, who may have experienced comorbid conditions and extensive treatments that make it difficult to attribute glial changes to the pain condition itself.

Given those challenges, in vivo imaging is seen as the best hope for exploring glial activation in humans. A number of researchers are pinning their hopes on positron emission tomography (PET) imaging of glial cells, but that will require molecular markers to specifically tag those cells. One study detected microglial activation in the thalamus, many years after limb amputation, using a ligand for the peripheral benzodiazepine receptoras a glial marker (Banati et al., 2001). Now, however, that marker is thought to be nonspecific, and could even be reflecting changes in neurons, says DeLeo. Thus, she says, her group and many others are avidly searching for new ways to tag specific cell populations.

Tracking how glia are involved in human pain will be a crucial next step for the field. Given the information that has already been amassed in animal models, finding roles for human glia in pain would likely unleash a store of fresh therapeutic targets.