Ballooned neurons in select neurodegenerative diseases contain phosphorylated neurofilament epitopes

Summary Ballooned neurons are histological features of several neurodegenerative diseases of the central nervous system. We describe the immunocytochemical staining of ballooned neurons in Pick's disease, unclassified dementia, corticonigral degeneration, pigment-spheroid degeneration and Alzheimer's disease. In all of these conditions the ballooned neurons contain phosphorylated epitopes recognized by monoclonal antibodies to neurofilaments, but not epitopes unique to Alzheimer neurofibrillary tangles and Pick bodies. The morphological features and immunohistochemical properties of ballooned neurons in these disorders bear resemblance to swollen neurons produced by neurotoxins that impair axoplasmic transport of neurofilaments. This finding, by analogy, suggests that impaired axoplasmic transport of neurofilaments may be a common mechanism in various dementing neurodegenerative diseases.     

Chromatolytic neurons in Werdnig-Hoffmann disease contain phosphorylated neurofilaments

Summary Two cases of Werdnig-Hoffman disease are described. Chromatolytic neurons were found within the ventral horns of the spinal cord as well as in dorsal root ganglia, Clarke's column, cranial nerve nuclei and thalamus. Immunohistochemical staining demonstrated ohosphorylated neurofilament antigen within the cytoplasm of the chromatolytic neurons. This finding suggests that a defect in slow axonal transport or in the regulation of neurofilament phosphorylation may play a role in the pathogenesis of this disorder.     

Axonal maturaion in development—I. Characterization of monoclonal antibodies reacting with axon-specific neurofilament epitopes

Monoclonal antibodies reacting with the high molecular weight neurofilament polypeptides (NF 150K and NF 200K) were obtained upon immunization with NF 150K and NF 200K isolated from bovine spinal cord by anion exchange chromatography. The five monoclonal antibodies obtained with NF 200K stained only axons. With three monoclonals the reactivity was abolished by digestion with phosphatase and by dilution of the supernatants in sodium potassium phosphate. The nine monoclonal antibodies obtained upon immunization with NF 150K stained both high molecular weight neurofilament polypeptides on immunoblots of bovine and rat spinal cord extracts with the exception of one monoclonal only reacting with the homologous antigen. The antibodies could be divided into two groups, axon-specific and conventional. Conventional antibodies decorated neurofilaments regardless of their location, i.e. axons, perikarya and dendrites. With all these antibodies the immunostaining was not affected by phosphatase digestion of neurofilament protein nor by dilution of the supernatants in sodium potassium phosphate. Axon-specific antibodies reacting with both NF 150K and NF 200K in rat spinal cord only stained the heterologous antigen (NF 200K) in rat optic nerve and sciatic nerve extracts. We suggest that some axon-specific neurofilament antibodies recognize neurofilament modifications other than phosphorylation; or, alternatively that they react with phosphorylated epitopes not accessible to phosphate or to exogenous phosphatases. Furthermore, we suggest that some neurofilament modifications do not occur uniformly throughout the nervous system.     

The effect of axotomy and deafferentation on phosphorylation dependent antigenicity of neurofilaments in rat superior cervical ganglion neurons

Previous immunocytochemical studies have shown immunological differences between neurofilaments in axons and those in dendrites and perikarya of many mature neuron types: it is now known that non-phosphorylated epitopes are normally seen in cell bodies and dendrites of mature neurons, whereas phosphorylated epitopes are observed in axons. Further studies on a variety of cell types have shown that phosphorylated epitopes are induced in the cell-body associated neurofilaments following axotomy, and we show here that comparable changes are observed in axotomized rat superior cervical ganglion neurons (SCG). We have also observed that preganglionic section induces similar changes in the levels of neurofilament phosphorylation, so that phosphorylated epitopes become visible in the cell bodies and dendrites of denervated neurons. The changes in this case appear more slowly and are accompanied by the appearance of punctate nuclear staining detectable with phosphorylation sensitive neurofilament antibodies. The acquisition of these staining patterns by both the axotomized and denervated SCG cells proved to be reversible, so that after 10–12 days no neurons in either experimental group exhibited perikaryal phosphorylated epitopes. These results indicate that changes in the level of neurofilament phosphorylation are not only associated with neuronal damage, but can also be induced by deafferentation.     

Phosphorylation-related immunoreactivity and the rate of transport of neurofilaments in chronic 2,5-hexanedione intoxication

Axonal transport of neurofilaments and the phosphorylation of epitopes on neurofilament proteins was studied in rats chronically intoxicated with 2,5-hexanedione. Sensory axons arising from the L₅ dorsal root ganglion exhibited accelerated transport and a reduced abundance of neurofilament proteins. The binding of an antibody to phosphorylated neurofilament epitopes was compared to the binding of an antibody to non-phosphorylated epitopes by quantitative ELISA. This immunochemical index of neurofilament phosphorylation was reduced in dorsal roots, proximal peripheral sensory axons and ventral roots, but not in a distal nerve (the nerve to the soleus). Axotomy produced a reduction in neurofilament protein abundance comparable to hexanedione without any change in the immunochemical phosphorulation index. These results are consistent with the hypothesis that the state of phosphorylation of neurofilaments in proximal axons is correlated with the rate of neurofilament transport.     

Normal neuronal cell bodies of the nucleus tractus mesencephalici nervi trigemini react with antibodies against phosphorylated epitopes on neurofilaments

Reactivity to antibodies directed against phosphorylated neurofilament epitopes is normally present in axons. Phosphorylated neurofilament immunoreactivity is not generally found in neuronal perikarya, except in abnormal states such as Alzheimer's disease. We found that cytoplasm of neurons of the nucleus tractus mesencephalici nervi trigemini in normal mice and rats reacts with monoclonal antibodies against phosphorylated epitopes on neurofilaments. This suggests either that phosphorylated epitopes on neurofilaments are localized in the perikarya of some normal neurons or that the antibodies that were used (SMI 31 and SMI 34) recognize more than phosphorylated epitopes.     

Neurite-like outgrowth from CNS explants may not always be of neuronal origin

In living explants of the developing CNS ‘true’ neurites (neuronal processes) may appear virtually indistinguishable from the processes of radial and Bergmann glia. This inability to clearly distinguish the phenotype of responsive ‘neurites’ in the living state has profound implications for the interpretation of responses to ‘neurite-promoting’ substances. Labeling techniques, specific for intermediate filament proteins that are characteristic of neurons and neuroglia, should always be used to supplement morphologic observations on neurite-like outgrowth from living CNS explants.     

Demonstration of hyperphosphorylated neurofilaments in neuronal perikarya in vivo by microinjection of antibodies into cultured spinal neurons

With conventional immunocytochemical techniques on fixed tissue, antibodies which recognize highly phosphorylated neurofilament proteins strongly label axons, but often react poorly with perikaryal neurofilaments. The reactivity of one such antibody, SMI31, with neurofilaments in vivo has been investigated by microinjecting purified SMI31 into large neurons in living cultures of embryonic mouse spinal cord. Microinjected SMI31 (SMI31I) labeled perikarya and dendrites in a fibrillar pattern indistinguishable from that of microinjected SMI32 (SMI32I), which labels hypophosphorylated neurofilaments of perikarya and dendrites in fixed tissue. SMI31 also labeled perikarya and dendrites when applied to whole unfixed cultures after extraction with 1% Triton X-100 or to cultures fixed in acetone after Triton-extraction, but prior to exposure to primary antibody. SMI31 labeled mainly axons when applied after fixation with acetone without Triton-extraction. Positive labeling of neurofilaments and various inclusions in neuronal somata with antibodies against highly phosphorylated neurofilaments has been described in a number of neurotoxic and neurodegenerative diseases and after neuronal injury. The results of this study indicate that explanations other than alterations in phosphorylation could account for these observations.     

Antibodies to the neuronal cytoskeleton are elicited by Alzheimer paired helical filament fractions

Antibodies were raised to paired helical filament (PHF) enriched fractions obtained from brains of individuals with Alzheimer disease by extraction with ionic detergent followed by sucrose gradient centrifugation. Electron microscopic examination showed that the fractions were enriched in Alzheimer PHF but contained also lipofuscin, amyloid, granular material and membranous elements. Analysis of these fractions with SDS-PAGE stained with Coomassie blue showed only a faint band at approximately 60 kDa while most of the material was excluded from the stacking gel. BALB/c mice were injected weekly with 100 or 200 μg of these fractions or corresponding fractions from age-matched control brains. The 3 mice injected with Alzheimer brain, but not the 5 mice injected with control brain fractions, produced antibodies that reacted with central and peripheral nervous system axons, Alzheimer neurofibrillary tangles in intact tissue as well as with isolated, SDS-treated paired helical filaments. In gel strips antibodies from all 3 mice injected with Alzheimer brain fractions reacted with the 200-kDa and 168-kDa but not the 68-kDa neurofilament subunits. The 3 antisera reacted also with some forms of the microtubule-associated protein τ. Adsorptions with the insoluble fraction from Alzheimer but not from control brains blocked staining of axons and NFT by all 3 antisera. Adsorption with highly purified neurofilament proteins or with a preparation containing the 200-kDa 168-kDa neurofilament subunits blocked axon and NFT immunostaining only in one antiserum. Adsorptions with microtubule protein, heat-stable microtubule-associated protein, or a preparation of τ did not completely block immunostaining by any of the 3 antisera. These results demonstrate that fractions enriched with Alzheimer paired helical filaments contain insoluble neurofilament, τ and other yet unidentified antigens.     

Neurofilament in feline primary afferent neurones: a quantitative immunocytochemical study

Two populations of cat L4 dorsal root ganglion (DRG) neurones were apparent from their Nissl staining with Toluidine blue. One had neurones of all sizes and the other had predominantly small neurones. The size distribution of neuronal profiles in the two populations overlapped and both were approximately normal. They corresponded to the light (L) and small dark (SD) cell populations previously described in rat DRGs. These neurones were examined with four antibodies to neurofilament: RT97, NFH, 155 and anti-68kD. RT97 is specific for the phosphorylated form of the 200 kDa subunit; NFH recognises both phosphorylated and non-phosphorylated forms of this subunit; 155 and anti-68 kDa recognise the 155 kDa and 68 kDa subunits respectively. The clearest differential labelling was seen with NFH and RT97 and this labelling was compared with cell size. High intensity NFH labelling was in population of neuronal profiles of all sizes and low intensity labelling in a population of predominantly small neuronal profiles. These populations corresponded respectively to the L and SD populations seen with toluidine blue staining. In the rat, these populations can be demonstrated by both NFH and RT97. In contrast in the cat, high intensity RT97 labelling was seen in only 75% of the L neuronal profiles defined with NFH and was also seen in some SD neuronal profiles defined with NFH. It is thus proposed that L and SD cell types are present in the cat DRG and can be demonstrated using the anti-neurofilament marker, NFH.     

Postnatal expressions of non-phosphorylated and phosphorylated neurofilament proteins in the rat hippocampus and the Timm-stained mossy fiber pathway

Neurofilament proteins (NFPs), the cytoskeletal proteins that are essential for axogenesis and maintenance of neuron shape in the nervous system, were studied for their spatial distributions at nine postnatal days (PN 3, 5, 7, 10, 14, 17, 21, 28, and 120). Simultaneously non-phosphorylated (SMI-32; 150/200 kDa; Sternberger) and phosphorylated (SMI-31; 200 kDa) NFP immunoreactivity in the entire developing rat hippocampus was studied, quantified, and compared to that of mossy fiber (MF) axons and terminals using Neo-Timm's histochemistry, the most selective, sensitive, and reproducible technique. Differential developmental expressions were observed between the two NFP states. SMI-32 was initially expressed on PN 3 only in the perikarya of pyramidal neurons in CA3. As early as PN 5, SMI-31 appeared in the MF pathway, in parallel to the growth of MF axons. By contrast, SMI-32 did not appear at any age in the MF pathway, including the MF terminal zone of stratum lucidum. At PN 14, the distribution of both NFPs in the MF system (MFs and their target neurons, i.e., CA3/CA4 pyramidal neurons and hilar neurons) was nearly complete; however, the peak densities of SMI-32 and SMI-31 were later at PN 21 and statistically equal to the most adult level (PN 120). The temporal regulation and maximal levels of SMI-32 and SMI-31 expressions on MF target neurons (CA3: SMI-32) and in the MF terminal zone (stratum lucidum: SMI-31) were nearly parallel to the progressive and rapid PN growth of the MF axons and terminals occurring between PN 14 and PN 17, suggesting that the mechanisms for maturation of MF synaptogenesis occur after PN 17.     

Ultrastructural observation on the thalamic neuronal inclusions in young mice

Summary Neuronal intracytoplasmic inclusions were found in the ventrolateral and posterior nuclei of the thalamus of 3-month-old thiamine-deficient mice and quaking mice.An electron-microscopic study revealed that these inclusions were composed of condensation of parallel rectilinear 10-nm filaments with interconnecting cross-bridges as a hallmark of neurofilaments. They were morphologically indistinguishable from those described in aged mice.This observation indicates that neuronal inclusions in the thalamus, which have been constantly reported in aging mice, can be seen even in young mice in the presence of cellular metabolic disturbances.     

Phosphorylated high molecular weight neurofilament protein in the peripheral motor,sensory and sympathetic neuronal perikarya: system-dependent normal variations and changes in amyotrophic lateral sclerosis and multiple system atrophy

Summary Using monoclonal antibody (Ta-51) that specifically binds phosphorylated high molecular weight neurofilament (pNFH) proteins, we investigated the occurrence of perikaryal pNFH in the spinal ventral horn motoneurons, intermediolateral column (ILC) neurons, sympathetic ganglion neurons and dorsal root ganglion (DRG) neurons obtained from patients with amyotrophic lateral sclerosis (ALS) and multiple system atrophy (MSA) and from control cases. In the controls, a system-dependent variation in perikaryal Ta-51 immunoreactivity was observed. Very few ventral horn cells and ILC neurons were stained with Ta-51, while large population of DRG neurons and sympathetic neurons were Ta-51 positive. The incidence of perikaryal immunoreactivity in the ventral horn cells was significantly increased in ALS and MSA. Some ILC neurons in ALS were Ta-51 positive and their incidence was significantly higher than that of the controls. These data suggest that both ILC neurons and ventral horn cells are affected with respect to pNFH metabolism in ALS and MSA. No significant difference was, however, detected in the Ta-51 immunoreactivity of both DRG and sympathetic ganglion neurons in ALS and MSA as compared with the controls.Supported by grants from the Japanese Ministry of Welfare and Health     

Effects of anesthesia on immunohistochemical detection of phosphorylated extracellular signal-regulated kinase in cerebral cortex

During attempts to examine the phosphorylation status of extracellular signal-regulated kinase (ERK) in cerebral cortex immunohistochemically, we determined whether deep anesthesia for euthanasia disturbs the phosphorylation status of ERK, because the anesthesia might influence activity-dependent phosphorylation of ERK. We compared effects of short (2 and 5 min) and long (>10 min) anesthesia by pentobarbital on the immunoreactivity for phosphorylated ERK in the visual and entorhinal cortices of rat. The long anesthesia drastically reduced the density of phosphorylated ERK immunopositive cells to about 15% of the short anesthesia condition. The reduction was observed in all cortical regions. We found no significant difference in pERK immunoreactivity between 2 and 5 min groups. A reduction of similar degree was induced by long anesthesia with isoflurane. Even if a similar duration of anesthesia is given, the immunohistochemical results possibly contain a variation due to the individual difference in the sensitivity to the anesthetics. We demonstrated that the variation of pERK immunopositive cell density in the visual cortex was significantly reduced by normalizing the values to the density in the nonvisual area in the entorhinal cortex, thus enabling us to detect differences between animals under different visual conditions with higher sensitivity. Therefore, the variation could be reduced by calculating the ratio of immunoreactivity in the area of interest to that in other cortical area as reference.     

Idiopathic Parkinson's disease: possible routes by which vulnerable neuronal types may be subject to neuroinvasion by an unknown pathogen

Summary. The progressive, neurodegenerative process underlying idiopathic Parkinson's disease is associated with the formation of proteinaceous inclusion bodies that involve a few susceptible neuronal types of the human nervous system. In the lower brain stem, the process begins in the dorsal motor nucleus of the vagus nerve and advances from there essentially upwards through susceptible regions of the medulla oblongata, pontine tegmentum, midbrain, and basal forebrain until it reaches the cerebral cortex. With time, multiple components of the autonomic, limbic, and motor systems become severely impaired. All of the vulnerable subcortical grays and cortical areas are closely interconnected. Incidental cases of idiopathic Parkinson's disease may show involvement of both the enteric nervous system and the dorsal motor nucleus of the vagus nerve. This observation, combined with the working hypothesis that the stereotypic topographic expansion pattern of the lesions may resemble that of a falling row of dominos, prompts the question whether the disorder might originate outside of the central nervous system, caused by a yet unidentified pathogen that is capable of passing the mucosal barrier of the gastrointestinal tract and, via postganglionic enteric neurons, entering the central nervous system along unmyelinated praeganglionic fibers generated from the visceromotor projection cells of the vagus nerve. By way of retrograde axonal and transneuronal transport, such a causative pathogen could reach selectively vulnerable subcortical nuclei and, unimpeded, gain access to the cerebral cortex. The here hypothesized mechanism offers one possible explanation for the sequential and apparently uninterrupted manner in which vulnerable brain regions, subcortical grays and cortical areas become involved in idiopathic Parkinson's disease. Received November 5, 2002; accepted December 3, 2002 Published online March 5, 2003 Acknowledgements This paper was supported by the Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie, the Deutsche Forschungsgemeinschaft (DFG), and Degussa (Hanau). The skillful technical assistance of Ms. I. Szász (graphics) is gratefully acknowledged. Author's address: Prof. H. Braak, M.D., Institute for Clinical Neuroanatomy, J.W. Goethe University, Theodor Stern Kai 7, D-60590 Frankfurt/Main, Germany, e-mail: Braak@em.uni-frankfurt.de     

HLA-DRB1*1501 risk association in multiple sclerosis may not be related to presentation of myelin epitopes

Susceptibility to multiple sclerosis (MS) is associated genetically with human leucocyte antigen (HLA) class II alleles, including DRB1*1501, DRB5*0101, and DQB1*0602, and it is possible that these alleles contribute to MS through an enhanced ability to present encephalitogenic myelin peptides to pathogenic T cells. HLA-DRB1*1502, which contains glycine instead of valine at position 86 of the P1 peptide-binding pocket, is apparently not genetically associated with MS. To identify possible differences between these alleles in their antigen-presenting function, we determined if T-cell responses to known DRB1*1501-restricted myelin peptides might be diminished or absent in transgenic (Tg) DRB1*1502-expressing mice. We found that Tg DRB1*1502 mice had moderate to strong T-cell responses to several myelin peptides with favorable DRB1*1501 binding motifs, notably myelin oligodendrocyte glycoprotein (MOG)-35-55 (which was also encephalitogenic), proteolipid protein (PLP)-95-116, and MOG-194-208, as well as other PLP and MOG peptides. These peptides, with the exception of MOG-194-208, were also immunogenic in healthy human donors expressing either DRB1*1502 or DRB1*1501. In contrast, the DRB1*1502 mice had weak or absent responses to peptides with unfavorable DRB1*1501 binding motifs. Overall, none of the DRB1*1501-restricted myelin peptides tested selectively lacked immunogenicity in association with DRB1*1502. These results indicate that the difference in risk association with MS of DRB1*1501 versus DRB1*1502 is not due to a lack of antigen presentation by DRB1*1502, at least for this set of myelin peptides, and suggest that other mechanisms involving DRB1*1501 may account for increased susceptibility to MS.     

Morphological studies on cerebral cortical lesions induced by transient ischemia in Mongolian gerbil —Diffuse and peripheral pallor of the neuronal perikarya

Summary Unilateral transient cerebral ischemia was produced in Mongolian gerbils by clipping the left common carotid artery for 1h. About 60% of the gerbils with neurological symptoms had post-ischemic seizures. The majority of those that had seizures died within a few days, and sections of their cerebral cortices contained many dark and shrunken neurons. However, the gerbils that did not have seizures survived without any severe complications. In the cerebral cortex of the latter, the neurons with diffuse or peripheral pallor of the perikarya were seen along with a small number of dark and shrunken neurons. Diffuse pallor occurred within a few hours following ischemia in layers III, V and VI, and disappeared 1 or 2 days after recirculation. Electron microscopically, these neurons showed dispersion of ribosomes, simple and elongated profiles of rough endoplasmic reticulum (r-ER), clustered vacuoles, and mild to moderate mitochondrial swelling. Occasional net-like tubulomembranous structures, probably derived from r-ER, were observed. On the other hand, peripheral pallor became apparent after 5 days following ischemia, usually involving layer II first and gradually extending to the deeper layers. Concomitantly, the amount of neuropil decreased and the dendrites exhibited tortuosity and irregularity in layer II. Electron microscopically, these neurons showed marked swelling of peripheral perikarya and polyribosomes and organelles were located peripherally to the nuclei. In addition, numerous degenerated axon terminals and distended dendrites were observed around the neurons. These observations indicate that diffuse pallor represents damage directly induced by ischemia and subsequent recirculation, while peripheral pallor is the delayed and remote effect of ischemia, probably due to degeneration of neuronal processes.     

CRMP2在大鼠海马神经元中的表达分布及对神经元突起生长的影响

目的探讨脑衰反应调节蛋白2(CRMP2)在大鼠海马神经元中的表达分布及对神经元突起生长的影响。方法CRMP2及其模拟磷酸化质粒CRMP2-S522D、CRMP2-T555D,转入体外培养的SD大鼠海马神经元,转染72 h后,采用免疫印迹法和免疫荧光染色分析CRMP2、CRMP2-S522D、CRMP2-T555D在神经元中的分布差异以及对神经元突起分枝和突起长度的影响。结果 CRMP2及CRMP2-T555D在神经元胞体和突起均有分布,CRMP2-S522D则主要分布于胞体。CRMP2-T555D对神经元的突起长度和分枝数均没有明显影响;CRMP2-S522D对突起数目没有明显影响,但可以显著降低突起长度。结论 CRMP2的不同磷酸化形式在大鼠海马神经元中存在分布差异,并对神经元的突起长度产生影响。     

Difference in pathogenesis between herpes simplex virus type 1 encephalitis and tick-borne encephalitis demonstrated by means of cerebrospinal fluid markers of glial and neuronal destruction

We determined the extent of neuronal and glial cell destruction in 13 patients with herpes simplex type 1 (HSV-1) encephalitis, 15 patients with tick-borne encephalitis (TBE), and 20 noninfectious controls by analyzing the cerebrospinal fluid (CSF) concentrations of neurofilament protein (a marker of neurons, mainly axons), neuron-specific enolase (a marker of neurons, mainly somas), glial fibrillary acidic protein, and S-100 protein (markers of astrocytes). In addition, in patients with HSV-1 encephalitis CSF samples were collected serially before 7, 8-14, and 18-49 days and 3-10 months after the onset of neurological symptoms. In the acute stage of HSV-1 encephalitis we found markedly higher CSF levels of the cell damage markers than in patients with TBE. The concentration of cell damage markers in HSV-1 encephalitis decreased within 45 days after acute infection, except for neurofilament protein. The CSF concentrations of neurofilament protein increased during the second week, remained extremely high throughout the next month, and decrease thereafter. The changes in these markers of neuronal and glial destruction demonstrate the neuronal and astroglial cell damage during the first month after HSV-1 encephalitis. In contrast, most patients with TBE had signs only of slight astrogliosis, except for two patients with paresis.     

Biphasic modulation in the trigeminal nociceptive neuronal responses by the intracerebroventricular prostaglandin E2 may be mediated through different EP receptors subtypes in rats

To determine which prostaglandin E₂ (PGE₂) receptor subtypes are involved in the brain-derived PGE₂-induced changes in nociception, we injected synthetic EP₁, EP₂ and EP₃ receptor agonists (0.01 fmol to 10 nmol) into the lateral cerebroventricle (LCV) of urethane-anesthetized rats and observed the changes in the responses of the wide dynamic range (WDR) neurons in the trigeminal nucleus caudalis to noxious pinching of facial skin. The enhancement and suppression of the nociceptive responses of the WDR neurons were observed after the LCV injection of MB28767 (an EP₃ receptor agonist) at a low dose range (1–100 fmol) and 17-phenyl-ω-trinor PGE₂ (an EP₁ receptor agonist) at high doses (1–10 nmol), respectively. Furthermore, the suppression of nociceptive neuronal responses after the LCV injection of PGE₂ (1 nmol) was completely blocked by SC19220 (an EP₁ receptor antagonist, 300 nmol). On the other hand, butaprost (an EP₂ receptor agonist) at any doses tested (0.1 fmol to 1 nmol) had no effect on the nociceptive responses. The LCV injection of MB28767 (10 fmol) and 17-phenyl-ω-trinor PGE₂ (1 nmol), which respectively enhanced and suppressed the nociceptive neuronal responses, did not affect the responses of the low threshold mechanoreceptive neurons to innocuous tactile stimuli. These results provide electrophysiological evidence that brain-derived PGE₂ induces mechanical hyperalgesia and hypoalgesia through EP₃ and EP₁ receptors, respectively, in the rat.