Spinal tolerance and dependence : Some observations on the role of spinal N-methyl-D-aspartate receptors and phosphorylation in the loss of opioid analgesic responses

The continuous delivery of opiates can lead to a reduction in analgesic effects. In humans, as in other animals, some component of this change in sensitivity seems likely to have a strong pharmacodynamic component. Such loss of effect, deemed to be tolerance in the present article, can be readily demonstrated in animals with repeated bolus and continuous intrathecal infusion of mu and delta opioids and alpha-2 adrenergic agonists. Research has shown that this loss of effect can be diminished by concurrent treatment with N-methyl-D-aspartate (NMDA) receptor antagonists and by the suppression of the activity of spinal protein kinase C (PKC). This suggests in part the probable role of PKC-mediated phosphorylation in the right shift in the dose-effect curves observed with continuous opiate or adrenergic exposure. Importantly, this right shift is seen to occur in parallel with an increase in the phosphorylating activity in the dorsal horn and in the expression of several PKC isozymes. The target of this phosphorylation is not certain. Phosphorylation of the NMDA receptor enhances its functionality, while phosphorylation of the opioid receptor or associated channels seems to diminish their activity or to enhance internalization. While the focus is on several specific components, the accumulating data emphasize the biological complexity of these changes in spinal drug reactivity.


ANALGESIC TOLERANCE
Antinociceptive effects produced by systemically (1)(2)(3) or spinally (4,5) delivered mu and delta opiates in animal models show a rapid decrement over an interval of hours to days.Spinal delivery of alpha-2 adrenergic agonists also produces antinociception.Chronic exposure results in a time-dependent decrement of activity (6).This loss of response to both classes of drug is characterized by a reduction in the effect produced by a given dose of that drug; an increase in the dose required to produce a given analgesic effect (eg, a shift of the dose-effect curve to the right); a decreasing maximum response (5,7,8); little cross-tolerance between agents that act at separate receptors (eg, mu versus delta [9]; mu versus alpha-2 [6,10]); an asymmetrical cross-tolerance between agents that are believed to act at the same receptor but differ in intrinsic activity (eg, sufentanil versus morphine [11], dexmedetomidine versus clonidine [12]); and after extended spinal exposure, significant hyperalgesia and spontaneous agitation evoked by withdrawal of the agonist by abstinence or by treatment with the respective antagonists (12)(13)(14).The signs of mu opiate withdrawal are suppressed by spinal alpha-2 agonists.
In humans, the issue of opiate analgesic tolerance is considerably more complex and not as well characterized as in the preclinical model.Even with extended intervals of spinal or systemic delivery in chronic pain patients, long intervals of stable drug requirement are observed (15)(16)(17).Nevertheless, moderate exposure to opiates postoperatively (18) or in cancer patients with stable pain over seven days (19) results in a modest loss of analgesic effect in humans.Spinal delivery of opiates to humans with chronic pain is often accompanied by an increase in the required dose over the period of infusion.In a retrospective examination of 163 cancer patients receiving continuous intrathecal infusion, an increase in the mean infusion concentration was noted (20).Inspection of the data revealed that some patients showed significant increases while many others did not.This is further emphasized when the same data are plotted as the median and 25th to 75th percentiles.The majority of patients showed no additional increase in the required dose after the first month.Plotting the individual increase in dosage in 33 patients over six months revealed that 21 displayed less than a threefold increase in starting dose.Thus, while many patients achieved protracted pain relief with spinal drugs, most patients showed some degree of dose increase.
Failure to see rapid changes in opiate consumption over time in clinically treated pain patients has been attributed to the possibility that the concurrent nature of the pain stimulus itself may alter tolerance development.Well defined preclinical studies have suggested that concurrent pain states (eg, inescapable pressure, arthritis or, in animals, an irritant in the paw) attenuate the loss of opiate responsiveness with chronic exposure (21)(22)(23)(24).In another study, loss of sensitivity to systemic opiates was enhanced by arthritic states or by stressors (25).The role of ongoing pain states in analgesic tolerance is thus surprisingly unclear, although interactions between afferent input and spinal biochemistry are evident.

POTENTIAL MECHANISMS UNDERLYING
DOSE ESCALATION Several factors may account for the loss of analgesic effect with continuous exposure and/or the increase in dose of drug required to produce a given magnitude of analgesia.

Change in pharmacokinetics
The abrupt return of pain in patients having achieved initial results from continuous spinal drug administration may result from changes in drug delivery or diffusion, including catheter removal; development of spinal cord compression; miscalculation of the dose; kinking or disconnection of catheters; development of a pseudomeningocele along the catheter or catheter investment, leading to a restricted spinal distribution of the injected drug; and change in drug spinal distribution secondary to catheter investment.These effects lead to an increase in the dose of agent required to produce a given degree of analgesia.

Change in pain intensity
Rapid increases in opiate dose requirements may reflect changes in the clinical state (eg, tumour metastasis and dysesthetic pain states).Alterations in the magnitude of the stimulus may result from changes in tumour mass, the presence of metastasis or development of pain states incidental to primary pathology (eg, bowel stasis, urinary retention, or septic bladder or kidney).Metastasis may lead to the distribution of disease to dermatomes that were previously unaffected (eg, fascia or bone), resulting in an enhanced activation of small afferents.Such increases predictably result in a rightward shift in the dose response curve for the analgesic agent (26)(27)(28)(29).Preclinical studies have indicated that the degree of right shift is inversely proportional to the intrinsic activity of the agonist (26).As stimulus intensity rises, the fraction of receptors that must be occupied is elevated and, theoretically, the intensity of the stimulus may be such that at near full receptor occupancy, the agent is unable to block transmission (ie, the agent technically becomes a partial agonist).

Change in pain mechanisms
Animal models have shown that different pain states may display a differential modulation by different classes of agents (30,31).Emergent pain states may result from evolving injuries to nerves (eg, developing tumours, radiation injury or chemotherapy).In such pain states, the effective stimulus may be the activation of low threshold mechanoreceptors (A-beta fibers) (32).Such input may activate systems that are not regulated in so efficacious a fashion by opiates as that input generated by small afferents (33).In systematic studies, the efficacy of epidural morphine was found to be as follows: somatic greater than visceral greater than radiating pain (34).Emergent neuropathies may thus result in the development of pain states that are less sensitive to opiates, though they may display sensitivity to other classes of agents, such as alpha-2 receptor agonists or N-type voltage-sensitive calcium channels (31).Such changes would serve to produce a right shift in the opiate dose effect relationship.

Change in psychosocial status
The evolution of pain states is frequently accompanied by significant changes in emotional status, the development of primary states of depression, and changes in lifestyle and coping mechanisms.Such alterations must be considered when significant changes in treatment efficacy are observed.

Change in pharmacodynamics (tolerance)
In well defined animal models, ongoing exposure to opiates results in a reduction in the responsiveness to the agent.Such changes occur in the absence of alterations in drug levels.While learning (ie, drug conditioning) can contribute to such changes, the loss of effect observed in chronic delivery studies occurs in the absence of prior training.This phenomenon in spinal drug delivery models is not secondary to decreased concentrations (eg, increased metabolism or clearance).Moreover, the phenomenon is often characterized by a right shift in the dose-effect curve with a decreasing maximum, a property suggestive of a reduction in the number of coupled receptors.

PHARMACODYNAMIC ISSUES
RELATED TO TOLERANCE As noted above, the continued exposure of the spinal cord to mu and delta opioid or alpha-2 adrenergic agonists (toleragen) results in a progressive loss of effect over time and a right shift in the dose-effect curve for a probe drug of the respective receptor class when assessed after extended exposure.While the mechanisms of this phenomenon are not understood, there is considerable evidence that the phenomenon may reflect complex changes in local system function.

SPINAL N-METHYL-D-ASPARTATE RECEPTOR:
TOLERANCE AND WITHDRAWAL One component of the change in system function relates to the original observation by Trujillo and Akil (35) that the loss of response to systemic opioids may be diminished by the concurrent delivery of antagonists for the N-methyl-D-aspartate (NMDA) glutamate receptors.The development of tolerance to mu opioid agonists delivered spinally (ie, reduced effect over time and/or a right shift of the probe dose-response curve) is diminished by a concurrent blockade of the spinal NMDA ionophore (36)(37)(38)(39).Although blockade of the spinal NMDA receptor alters tolerance, tolerance to spinally infused opioids occurs in the absence of increases in spinal glutamate release (14).This suggests that the role played by the spinal glutamatergic system reflects an enhanced responsiveness of the NMDA receptor and not an increase in release.
While spinal glutamate release was largely unaltered during the development of tolerance, it showed a large increase during withdrawal.This enhanced release correlated with the onset of hyperalgesia.Importantly, the hyperalgesia and agitation were reversed by the spinal delivery of NMDA antagonists and alpha-2 adrenoreceptor agonists.This apparent relationship between spinal withdrawal signs and glutamate release is consistent with the observation that spinal glutamate receptor activation induces agitation and a hyperalgesic state.
An interesting corollary to the above observations is that if activation of glutamate receptors facilitates tolerance, repeated withdrawal as produced by episodic naloxone administration would enhance the magnitude of tolerance because of increased periodic increases in glutamate release.Repetitive daily withdrawal led to an enhanced tolerance, and there was a progressive appearance of hyperalgesia and enhanced spinal glutamate release during the daily naloxone precipitated withdrawal (40).It should be noted that periodic withdrawal is an intrinsic component of studies in which bolus drug delivery is employed to provide opiate exposure.Such studies often report the evolution of a hyperalgesic state that is reversed by intrathecal NMDA receptor antagonism (39).Such hyperalgesia is not noted during the course of continuous spinal mu and alpha-2 infusion.

SPINAL SITES OF OPIATE AND ALPHA-2 ADRENERGIC ACTION DEMONSTRATING TOLERANCE
Opiates may act at both primary afferent sites and at nonprimary afferent sites in the dorsal horn.Accordingly, it is likely that inhibition of evoked small afferent transmitter release (eg, substance P [SP]) and the activation of dorsal horn neurons demonstrate a reduced inhibition with chronic spinal mu and alpha-2 agonist exposure.Chronic exposure of dorsal root ganglion cells results in the opiate effect being converted from an inhibitory effect to an excitatory effect mediated by Gs protein (41,42).This suggests a tolerance mediated at the level of the primary afferent.It is hypothesized that during withdrawal, there is a corresponding increase in afferent terminal excitability, as measured by SP release.Although this has not been directly assessed, behavioural hyperalgesia during withdrawal is attenuated by intrathecal neurokinin-1 antagonists (43), consistent with the observation that intrathecal SP produces hyperalgesia (27).
In spinal release studies, small afferent input leads to an increase in spinal glutamate and prostaglandin-2 release, which is blocked by morphine (44).Because morphine has an action on afferent terminals, the locus of morphine action in this model is not certain.In contrast, intrathecal SP evokes spinal glutamate and prostaglandin release (45).Because of the association of the neurokinin-1 site with postsynaptic sites (46), this release is hypothesized to be from nonprimary afferents.This direct 'postafferent' effect should also display a loss of effect with continued agonist exposure.Further, during withdrawal, it is likely that these transmitters display an increase in release.As noted above, previous work has demonstrated that there is an increase in spinal glutamate and prostaglandin-2 release during withdrawal (14).

SPINAL OPIATE AND ADRENERGIC RECEPTOR
ACTIONS AND PHOSPHORYLATION An important consequence of NMDA receptor activation is an increase in intracellular calcium and the activation of a variety of kinases.Protein kinase C (PKC) inhibitors diminish tolerance associated with spinal delivery of mu agonists and with approaches that prevent translocation of PKC (47,48).Spinal delivery of protein kinase A inhibitors in mice diminished the activity of morphine in tolerant animals (49,50).As with spinal nociceptive processing, the changes in spinal opiate sensitivity reflect a net balance of the constitutive activity of kinases and phosphatase (49,51).
There are several mechanisms by which a phosphorylation state may influence mu opiate action with chronic exposure.Mu opiate agonists and phorbol esters induce mu receptor desensitization, and this correlates with receptor phosphorylation (52).Importantly, phorbol ester, but not agonist-induced phosphorylation and desensitization, is blocked by the PKC inhibitor staurosporine (53).Thus, while consensus sites for PKC activity exist on the mu and alpha-2 receptor ( 52), the agonist-induced desensitization in at least one model is not mediated by PKC ( 54).An alternative route of phosphorylation is the G protein-coupled kinases.These enzymes can readily phosphorylate agonistoccupied (but not unoccupied or antagonist-occupied) receptors (55).Thus, inhibition of voltage-dependent calcium channels mediated by alpha-2 receptor desensitizes slowly with agonist exposure, and this desensitization is mediated by G protein-coupled kinases (56).Inhibition of calcineurin or activating PKC (both increasing phosphorylation) reduces pertussis toxin-sensitive G protein (G[PTX])-mediated inhibition of the N-type calcium channel (57,58).This effect appears to be selective because it is not seen for muscarinic receptor-mediated inhibition of the N-type calcium channel (mediated though a non-G[PTX] linkage) (58).Because the inhibitory effects of mu and alpha-2 receptors are coupled through PTX-sensitive linkage, these spinal receptors are expected to behave similarly.Given that phosphorylation diminishes the actions of the alpha-2 receptor, increased phosphorylating activity through an inositol triphosphate pathway by mu opiate receptor occupancy may provide an explanation as to why a modest cross-tolerance exists between mu and the alpha-2 receptor ( 6).Accordingly, inhibition of phosphorylation may prevent the modest crosstolerance otherwise observed.On the other hand, the link between the mu receptor and calcium channels is unlike the PKC-mediated linkages that couple mu receptor to NMDA channels in dorsal horn neurons (59,60) and the alpha-2 receptor to calcium channels in dorsal root ganglion neurons (61).For the mu receptor, a tight localization of the signalling pathway between receptor and channel was noted and hypothesized to reflect a direct link by a G protein (62).
An important intracellular linkage has been demonstrated between mu opiate and the NMDA receptor and mediated by phosphorylation (59,60).Mu opiates, through PKC, lead to an increased inositol triphosphate level (63,64).This coupling yields increased intracellular calcium derived from in-tracellular storage sites.Calcium leads to translocation of PKC to the plasma membrane.This leads to phosphorylation of the NMDA receptor, increasing the probability of channel openings and reducing the voltage-dependent magnesium block.This yields an enhanced NMDA response, which in turn leads to increased intracellular calcium.
A consequence of the above thinking is that in the presence of continued opiate exposure the cell is in a 'facilitated' state because of the increased effects associated with the phosphorylated NMDA receptor.Rohde and colleagues (65) demonstrated that the activation of spinal c-fos by formalin in the paw is exaggerated in morphine-tolerant animals.In recent work (71), we showed that on day 5 of morphine infusion, bolus intrathecal chelerythrine or GF109203X (GF), another PKC inhibitor, reversed in a dose-dependent fashion the loss of probe morphine effect otherwise observed when examined on day 6 (24 h after termination of the five-day morphine infusion sequence).The inactive homologue of GF, bisindolymaleimide V, was without effect (66).These results are consistent with the results of other laboratories that have reported that PKC inhibitors are able to diminish tolerance development (47,(67)(68)(69)(70)(71).These results suggest that chronic morphine infusion may increase spinal phosphorylating activity, and this increase is exacerbated by withdrawal.This likelihood is confirmed by our observation that on day six, after termination of intrathecal morphine infusion on day five, there was a twofold increase in dorsal horn, but not ventral horn, PKC phosphorylating activity.Importantly, the intrathecal dose of PKC inhibitor that reversed morphine tolerance was sufficient to block the increase in dorsal horn PKC-mediated phosphorylating activity (as measured by phosphorylation of a PKC-selective substrate).Importantly, these effects were not noted after an acute bolus of intrathecal morphine.Using 3 H-phorbol-12,13-dibutyrate binding, Mayer and colleagues (66) showed that daily spinal injection of morphine leads to an increase in membrane-bound PKC, particularly in spinal laminae I and II.
These results jointly suggest that, during opiate exposure, the change in spinal responsiveness may reflect the activation of spinal PKC.Of particular interest, not only is there an increase in kinase activation (as suggested by the increased phosphorylating activity) and the appearance of membranebound PKC, but an increased expression of PKCalpha/gamma was also observed, as measured by Western blots (66).This suggests an upregulation of phosphorylating capacity 24 h after termination of opiate infusion.These results suggest that, during the course of opiate exposure, activation of PKC is in part responsible for the loss of effect that occurs with continuous morphine exposure.

PHOSPHORYLATING ENZYMES
RELEVANT TO TOLERANCE Several PKC isozymes have been described.Which ones are functionally relevant to nociceptive processing is a subject of considerable interest.As noted above, morphine infusion resulted in an increased dorsal horn expression of both PKCalpha/gamma.Mao and colleagues (47) reported that daily Pain Res Manage Vol 5 No 1 Spring 2000 Yaksh and Hua repeated spinal morphine administration in rats is associated with an increase in the membrane bound PKC-gamma isoform.This isozyme is located in populations of neurons in the substantia gelatinosa, and knockout mice studies have emphasized the importance of this isozyme to the development of hyperalgesia (72), a phenomenon that also occurs in withdrawal.Nevertheless, we suspect that the increased phosphorylating activity noted above likely reflects upon an activation of a variety of PKC isoforms.In this regard, it should be emphasized that there is only a modest (ie, 5%) overlap between neurons expressing mu opioid receptors and PKC-gamma (73).It thus seems unlikely that the entire phenomenon of tolerance could be associated with that specific population of neurons (although they might mediate some or all of the hyperpathia).

ROLE OF SPINAL PHOSPHORYLATION
The mechanisms whereby phosphorylation alters spinal tolerance are likely to be complicated.Based on the data reviewed in the preceding sections, two broad classes of alternatives are considered.

Changes in nonopiate receptor function
Activation of kinases results in the phosphorylation of numerous membrane proteins that constitute receptor channels and enzymes that regulate neuronal excitability.One example of the effect of such phosphorylation is the assigned importance of the NMDA ionophore to the tolerance process.The studies with NMDA antagonists outlined above emphasize the importance of the NMDA receptor in spinal opiate tolerance.Phosphorylation of the NMDA receptor increases the functionality of the NMDA ionophore, which, accordingly, serves to increase calcium flux (74).This positive feedback would thus serve to enhance opiate receptor phosphorylation, with an attendant decrease in mu receptor function.Not incidentally, phosphorylation of the NMDA receptor appears to play an important role in the development of hyperalgesia (see above).

Opiate receptor-effector coupling
Preservation of the analgesic effect of chronic spinal morphine by PKC inhibition may reflect changes in receptoreffector coupling.Mu opioid receptors inhibit adenylyl cyclase activity through Gi, a glutamyl transpeptidase membrane-bound protein, decreasing cAMP formation (75).Decreased cAMP levels activate PKC, which then phosphorylates the mu receptor-coupled G protein.This linkage would thereby serve to suppress the receptor-evoked inhibition of adenylate cyclase (76,77).G protein-coupled receptor kinases may specifically phosphorylate and uncouple agonist-activated receptors and facilitate their interaction with beta-arrestins (53,78,79), by which the receptors may be further uncoupled from their G protein (78).

CONCLUSIONS
The present comments focus on the importance of the cascade that is initiated in part by the release of glutamate, the activation of local kinases and the ensuing phosphorylation of local components that alter the ability of the opiate occupancy of the receptor to diminish cellular excitability in the face of a high intensity afferent discharge (eg, nociception).As indicated, while we have focused on two players, the NMDA receptor and PKC, the probable contribution of other excitatory transmitters (such as SP, purines and the like) and other phosphorylating enzymes are likely to prove to be of considerable importance.On the other hand, that the studies have shown the effects to be robust and are observed to be associated with parallel results observed with alpha-2 agonists acting at another G-coupled protein receptor provides some sense that this cascade may contribute broadly to the overall mechanisms by which such downregulation in pharmacological effect is observed.Pain Res Manage Vol 5 No 1 Spring 2000