Peripheral nerves serve as pain's first messengers, firing action potentials as far as three feet to the spinal cord, where they alert the next nerve en route to the brain. Tissue damage and inflammation can lower nociceptor thresholds and increase neuronal excitability. The result is heightened pain sensitivity, a protective device, which resolves when the tissue heals.

But if something physically damages the peripheral nerve, the message is much bigger and more complex, with a flurry of activity at the site of injury, in the dorsal root ganglia – where the nerve's cell body is, and in the dorsal horn of the spinal cord – where the first synapse is located. Retrograde signals from the site of injury to the cell body affect transcription and alter expression of nociceptors and ion channels in peripheral nerves. Secondary pain transmission neurons in the spinal cord react to increased peripheral nerve activity with...


All three major MAP kinases, p38, ERK, and JNK, are activated in peripheral nerves after experimentally induced injury. In the spinal cord, p38 and ERK activation is also found, but in different cell types. Here, ERK is stimulated in neurons, and several groups have shown that p38 is stimulated solely in microglia, the resident immune cells of the nervous system. ERK is activated with acute painful stimuli, but not with nonpainful stimuli. P38 only shows up after injury.

P38, says Woolf, "is a key integrator of stress to cells. The stress may be direct, as to the dorsal root ganglia cells when they are injured. It may also involve stress to neurons in the dorsal horn due to massive activation." The dorsal horn is of particular interest to researchers, as it represents the initial processing site for pain. There, Woolf says, p38 activation in microglia is "such a prominent feature in first week or so, it swamps all others." After that, MAP kinases begin to be expressed in different cell types, perhaps reflecting the fact that there is an ongoing abnormal situation in the spinal cord, he says.

In a typical rat model of neuropathic pain, a ligature is tied tightly around a nerve. Within one day, animals withdraw their affected hind limb from the same gentle touch that was ignored prior to nerve injury. This touch-evoked response lasts for up to 10 weeks. In addition, rats exhibit behaviors consistent with spontaneous pain, including frequent licking of the affected hind paw.

Using p38 inhibitors, Woolf's group showed that not only is the enzyme activated, but it also contributes to pain hypersensitivity. Phosphorylated p38, the active form of the kinase, was assayed immunohistochemically at different time points after ligation of the fifth lumbar (L5) nerve. In the corresponding L5 segment of the spinal cord, p38 is activated within one day, but in the L5 dorsal root ganglia, where the nerve cell bodies are, p38 activation doesn't show up until day three. When p38 activity is blocked with a spinally infused inhibitor, rats show less allodynia for up to 10 days after nerve injury.1



Pain signals cause release of substance P and excitatory amino acids (EAAs). Activation of NK-1 receptors by substance P and AMPA receptors by EAAs cause transient depolarization of the pain transmission neurons (PTNs). NMDA-linked channels are normally inoperative as they are chronically plugged by Mg2+.

In the classic view of pathological pain (A), glia play no role and the intense and/or prolonged barrage of substance P and EAAs sensitizes the PTN. The PTN is depolarized such that the Mg2+ exits the NMDA-linked channel. The resulting influx of Ca2+ activates constitutively expressed nitric oxide synthase (cNOS) causing conversion of L-arginine to nitric oxide.

Increasing evidence, however, suggest that glia (B) become activated and are a driving force for creating and maintaining pathological pain states. Following activation microglia and astrocytes cause PTN hyperexcitability and exaggerated release of substance P and EAAs from presynaptic termini. (Adapted from L.R. Watkins et al., Trends Neurosci, 24:450–5, 2001.)

Linda Sorkin, professor of anesthesiology at the University of California, San Diego, studies tumor necrosis factor (TNF), which activates p38 in vitro. Using the same neuropathic pain model as Woolf, spinal nerve ligation, she shows that a TNF inhibitor blocks both the phosphorylation of p38 in the dorsal root ganglia and the development of the hypersensitive pain state.2 But timing is important. If the anti-TNF treatment is started even one day after nerve injury, it's ineffective. Thus, Sorkin says she believes she is studying the initiating sequence of events that convert nerve injury to chronic pain.

With regard to what turns p38 on in the spinal cord, "the search is on," says Linda Watkins, professor of psychology at the University of Colorado, Boulder. Imagine the peripheral neuron firing with intensity, sending signals into the dorsal horn of the spinal cord. "Some are very unique neuron-to-glia signals, such as fractalkine," says Watkins. Fractalkine is a protein in the chemokine family that is expressed only on the outside surface of neurons. Shed from neurons sending distress signals, fractalkines bind to and activate nearby glial cells. "The only cells that express receptors for [fractalkine] are microglia," says Watkins.

In microglia, p38 promotes the synthesis and release of proinflammatory cytokines, including interleukin (IL)-1, IL-6, and TNF, says Watkins. It leads to new transcription and stabilizes the mRNA, "so you get more translation for longer period of time," she says. Moreover, IL-1 is not made in an active form, but as a procytokine. "P38 activates IL-1-converting enzyme, which then activates IL-1 and leads to its release," explains Watkins. Thus, p38 performs a triple whammy on cytokine production, boosting transcription, translation, and posttranslational modification.

In accord with this scenario, Watkins and her group have shown that a p38 inhibitor decreased pain hypersensitivity in a similar fashion to antagonists of the proinflammatory cytokines.3


The response of spinal cord microglia to nerve injury constitutes more than phosphorylation of p38. One of three types of nonneuronal cells in the central nervous system (astrocytes and oligodendrites being the others), microglia are called the macrophages of the system. When quiescent, they have processes that sense the environment, says Joyce DeLeo, professor of anesthesiology and of pharmacology and toxicology at Dartmouth Medical School. When they become activated, she says, "The cell body gets larger, the surface antigen expression gets greater, and depending on if there's debris, they actually become phagocytic."

In addition to hypertrophy, there is hyperplasia. "Perhaps the most dramatic thing that [microglia] do is they divide," says Wolfgang "Jake" Streit, professor of neuroscience at University of Florida's Evelyn F. and William L. McKnight Brain Institute in Gainesville. "They divide like you wouldn't believe."

Rather than inhibiting signal transduction, DeLeo inhibits microglial activation with the antibiotic minocycline. Its mechanism is not entirely clear, but the inhibition is specific to microglia, leaving astrocytes and neurons unaffected. Not only does minocycline block the development of pain hypersensitivity, a related suppression of proinflammatory cytokines is observed.4

In DeLeo's study, the microglia inhibitor prevented the development of pain when treatment began prior to injury. But, similar to other studies, the inhibitor was not able to reverse increased pain when treatment came after the injury. This is an important distinction that may provide clues to the sequence of events in a progressive disorder. "Microglia become activated within minutes after nerve injury," says DeLeo. "Then something happens in the five to seven days after injury."

Addressing that something, DeLeo's working hypothesis is that astrocytes act to maintain the changes initiated by microglia. One job of astrocytes is removing glutamate from dorsal horn synapses via transporters. "The astrocyte is right there," says DeLeo.


Woolf's study showed that a specific p38 inhibitor could both prevent and reverse pain hypersensitivity, even when administered 10 days after nerve injury. To explain the discrepancy between his result and DeLeo's, Woolf says, "Our work indicates that beyond this initial microglial activation phase, there are other cells which begin to express p38." Current work in his lab seeks to identify these other cells. In any case, targeting enzyme rather than the cell type allows later stages of the progression to be interfered with, "when the microglial activation is no longer a prominent driver of the neuropathic pain," he says.

To complicate matters however, Sorkin was unable to reverse pain hypersensitivity using the same inhibitor and same nerve injury model as Woolf, a result, she jokes, that kept her up at night. "That totally perplexed me," says Sorkin. "There's always going to be discrepancies from lab to lab just because there are variations in technique," she says. She noticed that Woolf's group used older rats. With the spinal nerve ligation model, she says, the younger the rats, the more pain behavior was noted. "It's a peculiarity of that model."

Sorkin is currently testing whether differences in degree of pain sensitivity, based on differences in age or in the severity of the injury, affect how a treatment might have an effect on pain. She hypothesizes that with a "bigger burst of p38 activation right away, post-treatment doesn't work. Everybody agrees that pre-treatment works," says Sorkin. "It's just a matter of understanding under what conditions post-treatment works."

Obviously, reversal of neuropathic pain is a desirable capability for a potential medication. There may be situations in which treatment could be given when neuropathy is likely to develop, like surgery or chemotherapy. But for the most part, neuropathic pain may not even be diagnosed until months or years have passed and the injury has long resolved, says Frank Porreca, professor of pharmacology and anesthesiology at the University of Arizona in Tucson.

Differences between the animal models and clinical neuropathic pain pose pitfalls for those interested in translating their research. In people, Porreca says, "It's not even one condition, because patients suffer from really different symptoms. Most neuropathic pain patients suffer from spontaneous burning pain. A subset of those patients suffers from touch-evoked pain – allodynia." At present, animal models only measure evoked pain, not spontaneous pain. The two symptoms may not share the same mechanism.

A second point, Porreca says, is that the time scale studied in animals is relatively brief, up to 10 days after injury. Contrast that to the long-term condition in humans and anecdotal evidence that the condition is continually evolving, as when an effective medication suddenly stops working. "These are the uncertainties that we have to be aware of," says Porreca.

Interested in reading more?

Become a Member of

Receive full access to more than 35 years of archives, as well as TS Digest, digital editions of The Scientist, feature stories, and much more!
Already a member?