Lysophosphatidic acid drives to mirror-image pain via corpus callosum-mediated propagation of inflammatory responses

Abstract

Pain transmission is considered a unilateral process; however, development of bilateral hyperalgesia, including mirror-image pain (MIP), challenges this notion. The neural basis underlying bilateral hyperalgesia remains unclear. We investigated whether microglial activation within the corpus callosum is associated with interhemispheric propagation of inflammatory responses contributing to MIP. In a photothrombotic reperfusion mouse model, lysophosphatidic acid (LPA) signalling was associated with microglial activation in the contralateral corpus callosum and MIP development. Furthermore, PF8380 or minocycline suppressed bilateral hyperalgesia with MIP, supporting the therapeutic relevance of targeting LPA signalling and microglial activation. Together, these data support a model of an inflammatory circuit involving ischemic-core LPA, callosal microglia, contralateral insular PGE₂ signalling, and contralateral ACC activation that contribute to MIP. Because dysregulated LPA signalling and microglial activation are features of other neuroinflammatory conditions, transcallosal glial crosstalk may also contribute to widespread pain syndromes. Our findings highlight the therapeutic potential of targeting LPA-related pathways.

Lysophosphatidic acid (LPA) drives stroke-induced mirror-image pain by propagating inflammatory signals through the corpus callosum, involving LPA amplification, microglial activation, and increased PGE₂ production in the contralateral cortex.

Introduction

Stroke affects approximately 80 million people worldwide and remains a leading cause of death and long-term disability¹. Among its diverse neurological sequelae—ranging from motor and cognitive impairments to dysphagia and aphasia—central post-stroke pain (CPSP) is particularly challenging to treat and significantly impairs patients’ quality of life². While CPSP typically arises contralateral to the cerebral lesion^2,3, a subset of patients experiences mirror-image pain (MIP), characterised by hyperalgesia on the ipsilateral side^4,5.

MIP has been reported in other pain conditions, including neuropathic pain^6,7, complex regional pain syndrome⁸, and experimental models of inflammatory and peripheral nerve injury^9,10. The occurrence of MIP extends beyond the conventional understanding of unilateral pain processing^3,11, underscoring the urgent need to elucidate its underlying neural mechanisms. Despite its clinical relevance, the central and peripheral neural circuits that govern bilateral pain transmission remain largely unknown. Emerging evidence suggests that interhemispheric communication via the corpus callosum may mediate bilateral hyperalgesia¹⁰. Mechanisms such as astrocyte-driven cortical signalling and circuits involving the anterior cingulate cortex (ACC) have been implicated in this process^10,12,13.

We previously developed a long-lasting CPSP model¹⁴ by combining photochemically induced thrombosis (PIT)¹⁵ with tissue plasminogen activator (tPA), in which we observed MIP¹⁴. A key previous finding was that development of MIP was abolished in lysophosphatidic acid receptor 1 (LPAR₁)-deficient mice¹⁴. However, the neuro-pathophysiological mechanism by which LPA promotes MIP remains unknown. Our previous studies have suggested that LPA activates microglia and that a positive feedback mechanism exists whereby LPA promotes its own production^16–19. We aimed to identify potential therapeutic targets within the LPA–microglia axis. To this end, we employed advanced imaging mass spectrometry to investigate whether corpus callosum-associated propagation of inflammatory responses involving LPA is linked to microglial activation and MIP.

Results

Autotaxin (ATX)-dependent LPA signalling contributes to Ischaemic-reperfusion (IR) associated bilateral hyperalgesia with MIP

First, we examined whether IR after PIT induces bilateral hyperalgesia, as reported previously¹⁴. Pain thresholds were measured on day 3 after PIT targeting the left middle cerebral artery territory. Neither PIT nor tPA alone affected the mechanical pain threshold, whereas PIT followed by tPA administration 6 h later significantly decreased thresholds in both paws, including the ipsilesional paw showing MIP, compared with vehicle-treated controls (Fig. 1A-C). Throughout this manuscript, we refer to PIT+tPA-treated mice as the CPSP model.

Fig. 1. IR is associated with increased LPA in the cerebral cortex and the corpus callosum.

A Experimental timeline for reperfusion and pain testing in the CPSP model. The figure was created in BioRender. NEYAMA, H. (2026) https://BioRender.com/7g4nbpd. B, C Mechanical hyperalgesia in the left paw (B) and right paw (C) induced by IR. Comparative DESI-IMS images showing 18:1-LPA (D), PGD₂ (E), and PGE₂ (F) in sham + tPA and PIT + tPA mice on day 3. Scale bar: 1 mm. G Experimental timeline for ATX inhibitor (PF8380) treatment and pain testing in the CPSP model. PF8380 suppresses IR-induced mechanical hyperalgesia: (H) left paw; (I) right paw. Comparative DESI-IMS images on day 3 showing the effects of PF8380 on 18:1-LPA (J), PGD₂ (K), and PGE₂ (L). Scale bar: 1 mm. M, N Representative immunostaining images for Iba1 and CD68 in Veh- and PF8380-treated mice. Scale bar: 500 μm. Data are expressed as the mean ± S.E.M. B, C ****P < 0.0001. H, I *P < 0.05, ****P < 0.0001. Two-way ANOVA followed by Bonferroni’s multiple-comparisons test (n = 6 for B, C; n = 5 for H, I). Abbreviations: ATX autotaxin, CPSP central post-stroke pain, IR ischaemia-reperfusion, LPA lysophosphatidic acid, 18:1-LPA oleoyl lysophosphatidic acid, MIP mirror-image pain, PIT photo-induced thrombosis, tPA tissue plasminogen activator, DESI-IMS desorption electrospray ionisation imaging mass spectrometry, PGE₂ prostaglandin E₂, Iba1 ionised calcium-binding adaptor molecule 1, ANOVA analysis of variance, Veh vehicle.

We next hypothesised that LPA, a lipid signalling molecule produced in the ischaemic core, might initiate this cascade. Desorption electrospray ionisation imaging mass spectrometry (DESI-IMS) performed on day 3 in the CPSP model revealed a robust increase in the bioactive lipid oleoyl-LPA (18:1-LPA) within the infarct core. Notably, 18:1-LPA was also detected along the corpus callosum and in the contralateral cortex (Fig. 1D, white arrows).

Additionally, levels of prostaglandin D₂ (PGD₂) and prostaglandin E₂ (PGE₂), key inflammatory mediators, were elevated in the ischaemic core. The distinct spatial distributions are consistent with successful differentiation of PGD₂ and PGE₂. PGD₂ was characteristically distributed in the ischaemic penumbra surrounding the lesion (Fig. 1E), whereas PGE₂ showed a modest increase in the contralateral cortex (Fig. 1F, white arrows).

To investigate whether LPA is produced via ATX, the LPA synthesis enzyme, we performed behavioural assays assessing thermal (Hargreaves) and mechanical (digital von Frey) withdrawal responses. As shown in Fig. 1G and Supplementary Fig. 1A, the ATX inhibitor PF8380 (15 nmol/5 μL, 5 μL) was administered via lateral ventricle infusion 30 min before PIT treatment. PF8380 significantly reduced both mechanical and thermal hyperalgesia in both paws without causing locomotor dysfunction (Fig. 1H, I and Supplementary Fig. 1B–D). The absence of locomotor dysfunction supports the interpretation that the changes in withdrawal thresholds are not explained by gross locomotor impairment. DESI-IMS analysis indicated that 18:1-LPA production was suppressed by PF8380 not only in the ischaemic core but also in the corpus callosum (Fig. 1J, white arrows). Furthermore, PGD₂ and PGE₂ signals in the ischaemic core were also reduced following PF8380 administration (Fig. 1K, L). PF8380 additionally suppressed PGE₂ production in the contralateral cortex (Fig. 1L, white arrows).

To investigate the cellular source of LPA production, we next examined microglial involvement, based on previous reports that LPA activates microglia¹⁷. PF8380 reduced the number of immunostaining for ionised calcium-binding adaptor molecule 1 (Iba1)/Cluster of differentiation 68 (CD68) double-positive cells in the corpus callosum (Fig. 1M, N). These findings are consistent with a model in which LPA is associated with microglial activation and further endogenous LPA amplification in the corpus callosum.

Microglial activation in the corpus callosum contributes to MIP

To examine whether microglial activation in the corpus callosum is associated with LPA-related MIP, we performed (Iba1) on day 3 in the CPSP model. Dual immunostaining for Iba1 and CD68 in the corpus callosum revealed robust microglial activation in the PIT+tPA group but not in the sham+tPA group (Fig. 2Aa–e, Ba–e). In PIT+tPA mice, microglial activation was also observed in the contralateral cortex (Fig. 2Af–h, Bf–h).

Fig. 2. Association of MIP with microglial activation and LPA production in the corpus callosum.

Triple immunostaining for Iba1, CD68, and DAPI in Sham + tPA (A) and PIT + tPA (B). a Description of immunostaining. b Low-magnification image of immunostaining in the corpus callosum. Scale bar: 500 μm. High-magnification images of immunostaining in the corpus callosum: c Iba1, d CD68, and e merge. Scale bar: 100 μm. High-magnification images of immunostaining in the contralateral insular cortex: f Iba1, g CD68, and h merge. Scale bar: 50 μm C Experimental timeline for minocycline treatment and pain testing. Analgesic effect of minocycline (45 mg/kg, i.p.) on IR-induced mechanical hyperalgesia in the left paw (D) and the right paw (E). Comparative immunostaining and DESI-IMS images following minocycline treatment: F Iba1, G 18:1-LPA, H PGD₂, and I PGE₂. Scale bar: 500 μm (Iba1) and 1 mm (DESI-IMS). J Schematic diagram of AAV1-CD68-hM4-mCherry injection into the contralateral corpus callosum. K Experimental timeline for AAV injection into the corpus callosum and CNO treatment. Analgesic effect of CNO in AAV-injected mice on bilateral mechanical hyperalgesia in the left paw (L) and the right paw (M). N Schematic diagram of AAV1-CD68-hM4-mCherry injection into the contralateral insular cortex. O Experimental timeline for AAV injection into the contralateral insular cortex and CNO treatment. Analgesic effects of CNO in AAV-injected mice on bilateral mechanical hyperalgesia in the left paw (P) and the right paw (Q). Data are expressed as the mean ± S.E.M. D, E *P < 0.05, **P < 0.01, ***P < 0.0001; two-way RM ANOVA followed by Bonferroni’s multiple comparisons test (Veh n = 4, Minocycline n = 5). L, M ***P < 0.001, ****P < 0.0001; two-way RM ANOVA followed by Bonferroni’s multiple comparisons test (Saline n = 4, CNO n = 5). P, Q ****P < 0.0001; two-way RM ANOVA followed by Bonferroni’s multiple comparisons test (n = 7). Abbreviations: CC corpus callosum, MIP mirror-image pain, PIT photo-induced thrombosis, tPA tissue plasminogen activator, Iba1 ionised calcium-binding adapter molecule 1, CD68 cluster of differentiation 68, DAPI 4′,6-diamidino-2-phenylindole, AAV adeno-associated virus, CPSP central post-stroke pain, ANOVA analysis of variance.

These findings are consistent with sequential microglial activation associated with ATX-dependent LPA signalling and the development of MIP

Next, to clarify the mechanism linking microglial activation to MIP, we examined whether suppressing microglial function could prevent bilateral hyperalgesia, including MIP. Mice received minocycline (45 mg kg⁻¹, intraperitoneally), a microglial inhibitor^20,21, at five time points: 30 min before PIT, 30 min before tPA, and 30 min before each of the P1, P2, and P3 behavioural tests (Fig. 2C and Supplementary Fig. 2A). Vehicle-treated mice exhibited persistent bilateral hyperalgesia, including MIP in the left paw. In contrast, minocycline significantly attenuated both mechanical and thermal hyperalgesia on both sides without locomotor dysfunction (Fig. 2D, E and Supplementary Fig. 2B–D), indicating that microglial activation contributes critically to this behavioural phenotype. Furthermore, Iba1 immunostaining in the corpus callosum showed a robust reduction in microglial accumulation after minocycline treatment in PIT+tPA mice (Fig. 2F), and DESI-IMS indicated suppression of 18:1-LPA production in the corpus callosum (Fig. 2G, white arrows). Additionally, minocycline reduced PGE₂ and PGD₂ signals in the ischaemic core lesions (Fig. 2H, I).

Thus, to test whether local chemogenetic suppression of CD68 promoter-driven hM4-expressing cells in the contralateral callosal/cortical region affects MIP, we delivered AAV1-CD68-hM4Di-mCherry to the contralateral corpus callosum or insular cortex^22,23. Two weeks after viral injection, mCherry expression was confirmed by immunostaining. Among mCherry-positive cells, 51% (27/50 cells), 0% (0/51 cells), and 31% (20/65 cells) were co-localized with Iba1-, GFAP-, and NeuN-positive cells, respectively, indicating that transgene expression was present in both myeloid-lineage cells and neurons, but not in astrocytes (Supplementary Fig. 2E). Mice underwent PIT+tPA treatment and received clozapine-N-oxide (CNO; 10 mg kg⁻¹) at five time points: 30 min before PIT, 30 min before tPA, and 30 min before each of the P1, P2, and P3 behavioural tests (Fig. 2J, K, N, O). Under CNO treatment, hyperalgesia in the left paw (MIP) was abolished, without a comparable effect on the right paw response (Fig. 2L, M, P, Q; saline vs CNO).

These data indicate that chemogenetic modulation of AAV1-CD68-hM4Di-expressing cells in the contralateral callosal/cortical region attenuates MIP. Together with the minocycline and immunohistochemical data, these findings support an important contribution of contralateral microglia/myeloid responses, although the DREADD experiment alone cannot be interpreted as strictly microglia-specific.

Callosal proximity of regional LPA loading is associated with bilateral hyperalgesia and endogenous LPA responses

As shown in Fig. 1D, DESI-IMS analysis revealed that 18:1-LPA extended beyond the infarct core following IR, with signal observed along the corpus callosum and in the contralateral cortex. To test whether regional LPA loading near the infarct could reproduce this callosal pattern independently of ischemic injury, we stereotaxically injected 18:1-LPA (4 µg µL⁻¹, 1 µL) into the broad cortical region corresponding to the peri-infarct area in naïve, non-ischemic mice (Fig. 3A). Within 24 h, 18:1-LPA-injected mice showed significant bilateral mechanical hyperalgesia, whereas vehicle-injected mice did not (Fig. 3B). DESI-IMS of these brains showed 18:1-LPA signal along the corpus callosum (Fig. 3C, white arrows), consistent with a propagated endogenous response associated with callosal proximity.

Fig. 3. Callosal proximity of regional LPA injection is associated with bilateral hyperalgesia and endogenous LPA responses.

A Experimental schematic of 18:1-LPA (4 μg μL⁻¹, 1 μL) injection into the cerebral cortex at the indicated red region. B Bilateral reduction in pain thresholds following 18:1-LPA injection into the cerebral cortex. C DESI-IMS showing 18:1-LPA distribution in the corpus callosum after 18:1-LPA injection. Anatomical regions targeted for 18:1-LPA injection: D ischaemic core (IC) (cortex with corpus callosum [CC]); F IC (cortex without CC); H ischaemic penumbra (IP) (without CC); J IP (with CC). Corresponding changes in pain thresholds (mechanical stimulation test) for each injection site: E cortex with CC; G cortex without CC; I penumbra without CC; K penumbra with CC. L Anatomical region targeted for 18:1-LPA-d₅ injection. M DESI-IMS images showing the distribution of 18:1-LPA-d₅ in Veh- and 18:1-LPA-d₅-injected mice. N DESI-IMS images showing the distribution of 18:1-LPA in Veh- and 18:1-LPA-d₅-injected mice. Red dotted lines indicate ischaemic reference areas derived from the CPSP model. Data are expressed as the mean ± S.E.M. *P < 0.05, **P < 0.01, ****P < 0.0001; two-way ANOVA followed by Bonferroni’s multiple comparisons test. Sample sizes: n = 3 (B); n = 6 (E); n = 5 (G); n = 8 (I); n = 4 (K). Abbreviations: LPA lysophosphatidic acid, MS mass spectrometry, CC corpus callosum, CPSP central post-stroke pain, ANOVA analysis of variance, Veh vehicle.

To define the anatomical specificity underlying LPA-induced bilateral versus unilateral hyperalgesia, we performed region-specific microinjections of 18:1-LPA (4 µg µL⁻¹, 0.5 µL) followed by behavioural assessments. Injections into the insular cortex and primary somatosensory cortex (S1) proximal to the corpus callosum reliably evoked bilateral mechanical hyperalgesia (Fig. 3D, E). In contrast, injections into more distal portions of the insular cortex or into S1 regions farther from the callosal interface produced only contralateral hyperalgesia (Fig. 3F, G). LPA injections into cortical regions distant from the infarct core failed to induce hyperalgesia in either paw (Fig. 3H, I). Notably, direct injection of LPA into the corpus callosum yielded a distinct pain phenotype: ipsilateral hyperalgesia restricted to the side of injection (Fig. 3J, K). To test whether callosal proximity is associated with a contralateral callosal LPA response, we injected 18:1-LPA-d₅ into the somatosensory cortex while avoiding the corpus callosum, thereby allowing confirmation of the injection site. Local distribution of 18:1-LPA-d₅ was detected in 18:1-LPA-d₅–injected mice but not in vehicle-injected mice (Fig. 3L, M, yellow arrows). In contrast, no changes in 18:1-LPA distribution were observed in the corpus callosum of 18:1-LPA-d₅–injected mice relative to vehicle-injected mice (Fig. 3N). Additionally, local LPA injection into the somatosensory cortex did not induce microglial activation in the corpus callosum (Supplementary Fig. 3A–C).

These results suggest that callosal proximity is associated with propagated endogenous LPA production and bilateral hyperalgesia.

Callosal 18:1-LPA-d5 is associated with broader endogenous LPA responses and microglial activation

To determine whether local callosal LPA can evoke MIP-like behaviour and distinguish exogenous from endogenous LPA, we performed DESI-IMS using the deuterium-labelled tracer 18:1-LPA-d₅. To confirm that 18:1-LPA-d₅ retains biological activity comparable to native 18:1-LPA, we stereotaxically injected 18:1-LPA-d₅ (4 μg μL⁻¹, 0.5 μL) into the left corpus callosum of naïve mice and assessed mechanical hyperalgesia using digital von Frey testing 24 h later (Fig. 4A). Local administration significantly decreased the mechanical pain threshold in the left hind paw, consistent with an MIP-like phenotype, whereas vehicle injection had no effect (Fig. 4B, C). We next tested whether this behavioural effect depended on LPA₁ signalling by pre-treating mice with AM966, a selective LPAR₁ antagonist (1 nmol/5 μL, 5 μL, intracerebroventricularly), 10 min before LPA injection (Fig. 4D). AM966 prevented the MIP-like behaviour and restored pain thresholds in the left paw (Fig. 4E, F).

Fig. 4. Callosal 18:1-LPA-d₅ is associated with broader endogenous LPA responses and microglial activation in the corpus callosum.

A Experimental schematic of 18:1-LPA-d₅ (4 μg μL⁻¹, 0.5 μL) injection into the left corpus callosum. Decreased pain thresholds following 18:1-LPA-d₅ injection: B left paw, indicating MIP; C right paw. D Timeline for the AM966 (LPAR₁ antagonist) treatment experiment. AM966 suppresses 18:1-LPA-d₅-induced hyperalgesia: E left paw (MIP); F right paw. G DESI-IMS images showing 18:1-LPA-d₅ distribution after vehicle or 18:1-LPA-d₅ injection. H DESI-IMS images showing endogenous 18:1-LPA distribution after vehicle or 18:1-LPA-d₅ injection. Low-magnification immunostaining images showing Iba1 localisation in the corpus callosum after vehicle (I) or 18:1-LPA-d₅ (J) injection. Scale bar: 1 mm. K–V High-magnification immunostaining images at the white squares in (I, J): Iba1 (K, N), CD68 (L, O), CD11b (Q, T), GFAP (R, U), and NeuN (S, V) in vehicle- (K, L, Q–S) or 18:1-LPA-d₅-injected mice (N, O, T–V). Scale bar: 100 μm. M, P High-magnification double immunostaining images for Iba1 and CD68 at the white squares in I, J in vehicle- (M) or 18:1-LPA-d₅-injected mice (P). Scale bar: 100 μm. W Timeline for the minocycline treatment experiment. Inhibitory effects of minocycline on LPA-induced hyperalgesia: X left paw (MIP); Y right paw. Z DESI-IMS images showing endogenous 18:1-LPA distribution in vehicle- and minocycline-treated mice. Immunostaining for Iba1 (AA, AB) and GFAP (AC, AD) in the corpus callosum on day 1 post-injection: vehicle (AA, AC) and minocycline (AB, AD). Scale bar: 500 μm. Data are expressed as the mean ± S.E.M. (B, C, E, F, X, Y) ***P < 0.001, ****P < 0.0001, two-way ANOVA followed by Bonferroni’s multiple-comparisons test (n = 4–8). Abbreviations: MIP mirror-image pain, LPA lysophosphatidic acid, Iba1 ionised calcium-binding adaptor molecule 1, ANOVA analysis of variance.

To examine endogenous LPA responses after local tracer injection, we performed DESI-IMS analysis using the deuterium-labelled tracer 18:1-LPA-d₅. The 18:1-LPA-d₅ signal showed a localised distribution at the injection site (Fig. 4G). Endogenous 18:1-, 18:2-, 18:0-, and 20:4-LPA signals were also detected at the injection site (Fig. 4H and Supplementary Fig. 4A–C). Notably, 18:0- and 18:1-LPA showed a wider distribution within the corpus callosum in 18:1-LPA-d₅–injected mice than in vehicle-injected mice (Fig. 4H and Supplementary Fig. 4B). These findings support a broader increase in endogenous LPA within the corpus callosum, consistent with propagated local amplification rather than simple long-range diffusion of the injected tracer.

To clarify the involvement of microglia in the corpus callosum, we performed immunohistochemistry for Iba1. Iba1 immunoreactivity was broadly distributed throughout the corpus callosum following 18:1-LPA-d₅ injection but not in vehicle-treated mice (Fig. 4I, J). High-magnification images revealed that Iba1-positive cells induced by 18:1-LPA injection colocalised with CD68, a marker of activated microglia (Fig. 4K–P). In addition, CD11b, another microglial marker, showed morphological changes in 18:1-LPA-d₅–injected mice, further supporting microglial activation in response to LPA injection into the corpus callosum (Fig. 4Q, T). GFAP, an astrocytic marker, also showed activation in the corpus callosum (Fig. 4R, U) but not in the ACC (Supplementary Fig. 5A–F), whereas NeuN, a neuronal marker, showed minimal expression in the corpus callosum (Fig. 4S, V).

To determine whether LPA-induced hyperalgesia involves microglial activation, we performed behavioural assays using minocycline. Mice were administered minocycline (45 mg/kg, i.p.) 30 min before 18:1-LPA-d₅ injection (Fig. 4W). Systemic administration of minocycline prevented hyperalgesia in the left hind paw 1 day after LPA injection (Fig. 4X, Y). Furthermore, minocycline suppressed LPA production in the corpus callosum (Fig. 4Z, white arrows) and inhibited microglial activation (Fig. 4AA, AB), but did not inhibit astrocyte activation (Fig. 4AC, AD).

Callosal LPA contributes to MIP via contralateral PGE₂ production in the insular cortex

Building on prior findings that LPA induces PGE₂ production via microglial activation²⁴ and our observation in Fig. 1L, we hypothesised that corpus callosum–derived LPA promotes pain through PGE₂ production in the contralateral cerebral cortex.

DESI-IMS analysis revealed that arachidonic acid, a precursor of PGE₂, and PGE₂ were distributed in the bilateral ACC and the right insular cortex (Fig. 5A–C). To identify the cell types associated with PGE₂ production, we performed double immunostaining using neuronal and glial markers together with cyclooxygenase-2 (COX-2), a key enzyme in PGE₂ synthesis. Iba1 staining showed broad cortical distribution, including the insular cortex (Fig. 5D). CD11b and COX-2 were expressed in the contralateral insular cortex, and double staining revealed clear colocalisation of CD11b with COX-2 (Fig. 5E–G). In contrast, GFAP expression was relatively sparse in the contralateral insular cortex and showed little colocalisation with COX-2 (Fig. 5H–J). NeuN staining showed limited colocalisation with COX-2 (Fig. 5K–M). These findings suggest that CD11b-positive microglia/myeloid cells, and to a lesser extent neurons, are major candidate contributors to PGE₂ production in the right insular cortex. To determine whether microglial inhibition suppresses LPA-induced PGE₂ elevation in the contralateral insular cortex, we performed DESI-IMS analysis following minocycline treatment. As in Fig. 4W, minocycline was administered 30 min before 18:1-LPA-d₅ injection (Fig. 5N). DESI-IMS analysis showed that minocycline suppressed the elevation of PGE₂ in the contralateral insular cortex (Fig. 5O).

Fig. 5. PGE₂ signalling in the contralateral insular cortex is associated with MIP development.

A Experimental schematic of deuterated 18:1-LPA (18:1-LPA-d₅) injection into the left corpus callosum. DESI-IMS images on day 1 following 18:1-LPA-d₅ injection: B arachidonic acid distribution; C PGE₂ distribution. D Low-magnification image of Iba1 immunostaining in the contralateral cerebral cortex including insular cortex of 18:1-LPA-d₅-injected mice. Scale bar: 1 mm. E–M High-magnification immunostaining images: CD11b (E), GFAP (H), NeuN (K); COX-2 (F, I, L); merged images (G, J, M). Scale bar: 100 μm. N Experimental timeline for DESI-IMS analysis in minocycline-treated mice. O Comparative DESI-IMS images showing PGE₂ production in vehicle- and minocycline-treated mice. P Sites of PGE₂ injection into the contralateral right insular cortex. Changes in pain thresholds measured by mechanical stimulation on day 1 after PGE₂ administration into the right insular cortex: Q left paw; R right paw. S Timeline for the diclofenac (COX inhibitor) treatment experiment. Diclofenac inhibits 18:1-LPA-induced hyperalgesia: T left paw (MIP); U right paw. Data are expressed as the mean ± S.E.M. Q, R, T, U **P < 0.01, ***P < 0.001, ****P < 0.0001, two-way ANOVA followed by Bonferroni’s multiple-comparisons test (n = 3–7 for Q, R; n = 5 for T, U). Abbreviations: LPA lysophosphatidic acid, 18:1-LPA-d₅ deuterated 18:1-LPA, PGE₂ prostaglandin E₂, MIP mirror-image pain, MS mass spectrometry, COX cyclooxygenase, ANOVA analysis of variance.

To examine whether PGE₂ signalling contributes to LPA-induced hyperalgesia, we directly microinjected PGE₂ into the contralateral right insular cortex (Fig. 5P) and left corpus callosum (Supplementary Fig. 6A). Injection into the right insula and left corpus callosum induced hyperalgesia in the left hind paw but not in the right paw (Fig. 5Q, R and Supplementary Fig. 6B, C). These data support a contribution of PGE₂ signalling to the early expression of LPA-induced MIP. These findings are consistent with a model where callosal LPA is linked to contralateral insular PGE₂ signalling, rather than PGE₂ itself serving as the transcallosal propagating signal. To further test whether PGE₂ signalling participates in the early phase of this response, we administered the nonsteroidal anti-inflammatory drug diclofenac at three time points: 30 min before and 30 and 60 min after LPA injection (Fig. 5S). Diclofenac significantly reduced hyperalgesia in the left paw, consistent with the MIP phenotype (Fig. 5T, U), supporting a contribution of PGE₂ production to the early phase of LPA-induced MIP.

Taken together, these findings support a model in which callosal LPA is associated with microglia-linked neuroinflammation and PGE₂ signalling in the contralateral insular cortex, contributing to the development of MIP.

Contralateral insula–ACC axis orchestrates MIP

To investigate where MIP signals are ultimately processed, we performed c-Fos immunostaining to assess neuronal activation. We focused on the ACC, which is well recognised as a central hub in post-stroke pain²⁵. Direct microinjection of 18:1-LPA into the left insular cortex of naïve mice significantly increased c-Fos expression, a marker of neuronal activation, in both the ipsilateral and contralateral ACC compared with Veh injection (Fig. 6A–G).

Fig. 6. Involvement of the LPA–corpus callosum–contralateral insular–ACC pathway in MIP development.

c-Fos immunostaining of the ACC on day 1 following 18:1-LPA injection into the insular cortex: A Veh; D 18:1-LPA. Scale bar: 100 μm. B, E Double immunofluorescence images showing c-Fos (red) and DAPI (blue) in the left ACC: B Veh; E 18:1-LPA. Scale bar: 50 μm. C, F Representative images of c-Fos and DAPI staining in the right ACC: C Veh; F 18:1-LPA. Scale bar: 50 μm. G Quantification of c-Fos⁺ cells in the ACC following 18:1-LPA injection. Schematic diagrams showing ibotenic acid and 18:1-LPA injection sites: H left ACC; M right ACC. I, J, N, O NeuN immunostaining 1-week post-injection. Scale bar: 50 μm. I, N Veh; J, O Ibotenic acid. Effect of left ACC neuronal ablation on LPA-induced hyperalgesia: K left paw; L right paw. Effect of the right ACC neuronal ablation on LPA-induced hyperalgesia: P left paw; Q right paw. R Schematic of ibotenic acid and 18:1-LPA injection sites. NeuN immunostaining 1-week post-injection: (S) Veh; T Ibotenic acid. Scale bar: 50 μm. Effect of neuronal ablation in the contralateral insular cortex on 18:1-LPA-induced hyperalgesia: U left paw; V right paw. W Schematic diagram showing regions targeted for 18:1-LPA and minocycline administration. X Experimental timeline for minocycline (45 mg/kg, i.p.) and 18:1-LPA injection. Suppressive effects of minocycline on 18:1-LPA-induced bilateral mechanical hyperalgesia: Y left paw; Z right paw. AA Quantitative analysis of c-Fos expression in minocycline-treated mice at the bilateral ACC. c-Fos immunofluorescence images for vehicle (AB) or minocycline (AC) treatment in the left ACC. c-Fos immunofluorescence images for vehicle (AD) or minocycline (AE) treatment in the right ACC. Scale bar: 100 μm. Data are expressed as the mean ± S.E.M. (AB–AE). G *P < 0.05, **P < 0.01, unpaired t-test (n = 3–5). K, L, P, Q, U, V ***P < 0.001, ****P < 0.0001, two-way ANOVA followed by Bonferroni’s multiple-comparisons test (n = 6). Y, Z ****P < 0.0001, two-way RM ANOVA followed by Bonferroni’s multiple-comparisons test (n = 6). AA *P < 0.05, unpaired t-test (n = 6). Abbreviations: ACC anterior cingulate cortex, c-Fos immediate early gene marker of neuronal activity, DAPI 4′,6-diamidino-2-phenylindole, LPA lysophosphatidic acid, MIP mirror-image pain, NeuN neuronal nuclei, ANOVA analysis of variance, Veh vehicle.

To determine the pathway mediating transmission of LPA-induced bilateral hyperalgesia to the ACC, we selectively ablated each ACC using ibotenic acid 1 week prior to LPA injection. Ibotenic acid treatment markedly reduced the number of NeuN-positive neurons compared with the vehicle group (Fig. 6H–J, M–O, R–T). Although increased c-Fos expression was observed throughout the ACC (Fig. 6A, D and Supplementary Fig. 7Aa), ibotenic acid lesions were predominantly localised to the Cg2 region of the ACC (Supplementary Fig. 7Ab, c). Ablation of the left ACC selectively attenuated mechanical hyperalgesia in the right hind paw, whereas the left hind paw remained unaffected (Fig. 6K, L). Conversely, ablation of the right ACC selectively attenuated mechanical hyperalgesia in the left hind paw but not in the right hind paw (Fig. 6P, Q).

To further examine the role of the insular cortex, we surgically ablated the contralateral right insular cortex 1 week prior to LPA administration. Insular ablation was targeted to the medial insular cortex (Supplementary Fig. 7B). This intervention abolished hyperalgesia in the left paw, whereas the right paw remained responsive (Fig. 6R–V).

These findings suggest an asymmetric functional requirement of ACC nodes in this paradigm and support involvement of a corpus callosum–dependent route linking insular LPA to contralateral ACC activation during MIP.

To determine whether microglia contribute to neuronal activation in the ACC, we next performed behavioural and immunohistochemical analyses using the microglial inhibitor minocycline. Minocycline was administered 30 min prior to 18:1-LPA injection into the insular cortex (Fig. 6W, X). Minocycline significantly suppressed bilateral mechanical hyperalgesia (Fig. 6Y, Z). Notably, minocycline also significantly reduced c-Fos expression in the bilateral ACC (Fig. 6AA–AE). These findings are consistent with a contribution of minocycline-sensitive microglial responses to ACC activation in this paradigm.

Taken together, these results support a model involving an insular cortex–corpus callosum–contralateral insular cortex–ACC axis, in which insular LPA is associated with callosum-dependent signalling and contralateral ACC activation during MIP.

Working model of LPA-associated bilateral hyperalgesia involving corpus callosum microglia

This figure summarises a proposed working model for bilateral hyperalgesia after cerebral infarction. Typically, a left hemispheric infarct induces pain in the contralateral (right) lower limb (left panel). In cases of MIP, pain also emerges in the non-infarcted (left) limb, potentially progressing to bilateral hyperalgesia. Our data suggest that LPA accumulates in the infarcted cortex and is associated with broader inflammatory responses involving the corpus callosum (Fig. 1). Along this interhemispheric route, LPA-related signalling is associated with microglial activation in the corpus callosum and contralateral cortex.

Regional LPA administration near the corpus callosum was associated with increased callosal endogenous LPA responses, and pharmacologic inhibition of contralateral CD68-positive microglia/myeloid cells alleviated MIP (Figs. 2 and 3), supporting an important contribution of callosal and contralateral microglial responses to MIP.

Finally, LPA-associated neuroinflammation was linked to PGE₂ production in the contralateral insular cortex and to ACC activation (Figs. 5 and 6). Together, these findings support a model involving ischaemic-core LPA, callosal microglia, contralateral insular PGE₂ signalling, and ACC activation in bilateral hyperalgesia.

Discussion

In this study, we investigated the neuropathological mechanisms that may contribute to bilateral hyperalgesia with MIP. Our data support a model in which LPA contributes to MIP through corpus callosum-associated propagation of inflammatory responses. In addition, the combined pharmacological, imaging, and chemogenetic data support an important contribution of the LPA–microglia axis to this process. The observation that minocycline suppresses abnormal pain behaviour further suggests that modulation of microglial responses may have therapeutic relevance in central pain syndrome.

CPSP typically manifests as pain in the limbs contralateral to the infarcted region^2,3. This pattern aligns with the classical nociceptive pathway, wherein pain signals from peripheral tissues are transmitted to the spinal cord, decussate to the contralateral side, and ascend to the somatosensory cortex^3,11. However, in some cases, patients experience pain on the same side as the lesion, deviating from this canonical neural pathway. This atypical presentation, referred to MIP, is one cause of bilateral hyperalgesia^4,5. MIP has been reported in various conditions, including neuropathic pain^6,7, complex regional pain syndrome⁸, and experimental models of inflammation and peripheral nerve injury^9,10. The clinical relevance of systemic pain syndromes such as fibromyalgia further underscores the importance of understanding bilateral pain mechanism, although extrapolation of the present stroke model to other widespread pain conditions will require additional study.

Our experiments revealed abnormal microglial activation in the corpus callosum after ischaemic reperfusion, with additional activation observed in the contralateral cortex (Fig. 2B). Notably, inhibition of CD68-positive microglia/myeloid cells on the contralateral side significantly attenuated MIP (Fig. 2J–P). To achieve this local inhibition, we used an AAV vector carrying the hM4 receptor under the control of the CD68 promoter and applied DREADD technology. CNO administration, which activates hM4, has been reported to suppress microglial reactivity and pro-inflammatory cytokine expression (e.g., IL-1β and TNF-α) in LPS-induced models^22,23,26.

In this study, we employed an AAV1-CD68-hM4Di-mCherry construct with the aim of suppressing CD68-positive microglia/myeloid cells. This strategy was based on previous studies reporting CD68 promoter-driven inhibitory DREADD expression in spinal microglia and in the rat brain^22,23,26. However, because AAV-mediated targeting of microglia is technically challenging and context-dependent, we directly assessed mCherry expression in our preparation. In representative injection-site images, 54% of mCherry-positive cells were co-localized with Iba1-positive cells; no co-localization was observed with GFAP-positive cells, and 31% were co-localized with NeuN-positive cells. These findings indicate that the construct was expressed in Iba1-positive myeloid-lineage cells, but also showed partial neuronal expression (Supplementary Fig. 2E). Therefore, the behavioral effects observed after CNO administration cannot be attributed solely to microglia-specific inhibition, and a contribution of neuronal hM4Di activation cannot be excluded. Nevertheless, together with the minocycline experiments and Iba1/CD68 immunohistochemistry, these findings support an important contribution of microglia/myeloid responses to MIP. Future studies using more selective microglia-specific genetic approaches will be necessary to define the precise contribution of microglia.

Further studies using approaches with higher cell-type specificity will be necessary to clarify the precise contribution of microglia to the observed phenotypes.

In our study, CNO alleviated MIP, supporting an important contribution of contralateral CD68-positive microglia/myeloid cells to MIP. Furthermore, MIP induced by unilateral LPA injection into the cerebral cortex, including the corpus callosum, was abolished by neurotoxic ablation of the contralateral insular cortex (Fig. 6R–V), suggesting that contralateral microglia responses are linked to neuronal excitability in the contralateral insular cortex.

18:1-LPA signals were observed along the corpus callosum after peri-infarct cortical 18:1-LPA administration (Fig. 3C), whereas somatosensory cortical injection avoiding the corpus callosum did not alter callosal endogenous 18:1-LPA (Fig. 3N). In addition, DESI-IMS showed broader endogenous 18:1-LPA responses after callosal 18:1-LPA-d₅ injection despite retention of the tracer near the injection site (Fig. 4). Together, these findings are consistent with local amplification of endogenous LPA responses in the corpus callosum.

Our previous studies have shown that LPA stimulates microglial cytokine release, which in turn activates neuronal cPLA2, leading to further LPA synthesis, forming a self-amplifying loop^16–19. However, the corpus callosum contains few neuronal cell bodies (Fig. 4S, V) and is composed mainly of axonal fibres, making this neuron-dependent amplification mechanism unlikely to operate directly in this region. Alternatively, Liu et al.²⁷ reported that LPA enhances PLA2 activity in macrophages, promoting LPA synthesis via the LPC pathway. Importantly, the conversion of LPC to LPA requires the extracellular enzyme ATX²⁸, suggesting that intracellular LPA production in microglia is limited. In a cerebral ischaemia model, astrocytes have been shown to secrete excessive ATX²⁹. Thus, it is plausible that LPA production in the corpus callosum is facilitated by ATX derived from astrocytes acting on LPC produced by activated microglia. We also observed elevated PGE₂ levels in the contralateral insular cortex following LPA injection. LPA-induced hyperalgesia was significantly reduced by diclofenac (an NSAIDs), suggesting that PGE₂ signalling contributes to MIP development (Fig. 5T, U).

LPAR₁ is rarely expressed in neurons within the CNS³⁰, making it unlikely that LPA directly enhances neuronal activity. However, LPA has been reported to promote COX-2 expression and PGE₂ production in microglial systems^24,31,32. PGE₂ levels increase during brain injury^33–35 and mediate pain via the EP₃ receptor³⁶. Together with our data, these observations support a potential role for the LPA–PGE₂ axis in the pathogenesis of MIP. However, although PGE₂ injection into the left corpus callosum^33–35 induced hyperalgesia in the left paw (Supplementary Fig. 6A–C), consistent with MIP, DESI-IMS data did not demonstrate an elevation of PGE₂ in the corpus callosum (Figs. 1F and 5C). Therefore, PGE₂ is unlikely to function as a propagating signal through the corpus callosum. Instead, it should be interpreted more cautiously as a downstream effector or amplifier in cortical target regions.

Previous studies have emphasised the importance of interhemispheric communication via the corpus callosum in mediating MIP. Consistently, our results suggest that local increases of LPA in the corpus callosum are associated with microglial activation and broader endogenous LPA responses across hemispheres, contributing to bilateral mechanical hyperalgesia.

The corpus callosum is a major bundle of commissural fibers that plays a critical role in interhemispheric communication. However, its connectivity is not uniform; rather, it is highly region-specific^37,38. In particular, callosal projections are not limited to homotopic regions but can also extend to heterotopic cortical areas, including higher-order regions such as the insular cortex³⁹. Therefore, callosal function cannot be simply explained as symmetric information transfer between the two hemispheres.

In the present study, LPA administration into the left corpus callosum induced a mirror-image pain phenotype characterized by hypersensitivity predominantly on the left side. While this finding suggests that LPA signaling within the corpus callosum can trigger mirror-image pain, the mechanism underlying the apparent contralateral selectivity remains to be clarified.

In this context, our results suggest that callosal activation does not produce uniform bilateral cortical activation, but rather selectively recruits specific contralateral cortical networks. Such selectivity may arise from heterotopic projections together with spatially and temporally non-uniform callosal signaling^40,41.

Notably, we observed a selective increase in PGE₂ levels in the contralateral insular cortex following LPA administration into the left corpus callosum. This molecular asymmetry supports the idea that callosal activation preferentially engages specific contralateral cortical networks rather than inducing uniform bilateral responses. Given the well-established role of PGE₂ in pain sensitization, this finding may provide a mechanistic link between circuit-level selectivity, local inflammatory signaling, and the lateralized behavioral hypersensitivity observed in this study.

Regarding the final processing of nociceptive signals, our ablation findings suggest an asymmetric functional requirement of ACC nodes in this paradigm (Fig. 6). Selective ablation of neurons in either ACC suppressed mechanical hyperalgesia in the opposite paw, indicating that ACC activity on each side contributes to the corresponding behavioural output, likely in concert with insular cortex pathways. Notably, neuronal ablation was more pronounced in Cg2 than in Cg1 (Supplementary Fig. 7A), consistent with the engagement of the affective dimension of pain^42,43. Furthermore, ablation of the right insular cortex selectively improved left paw hyperalgesia, suggesting that inter-insular communication may also contribute (Fig. 7).

Fig. 7. Working model of bilateral hyperalgesia development.

Following left hemispheric stroke, hyperalgesia is typically induced in the contralateral hind paw (right paw; left panel). Aberrant LPA production in the left hemisphere propagates inflammation signaling to the contralateral hemisphere through corpus callosum microglia, leading to neuronal activation in the insular cortex and subsequent hyperalgesia in the ipsilateral hind paw (left paw; right panel). This mechanism may contribute to the development of mirror-image pain. Abbreviation: ACC anterior cingulate cortex, LPA lysophosphatidic acid, PGE₂ prostaglandin E₂.

The ACC is widely recognized as an integrative hub for both sensory and affective components of pain^44–46. Because of its role in emotional processing, it has often been considered to exhibit weak or absent contralateral bias compared to primary sensory cortex^46,47. However, in the present study, we observed clear contralateral bias in ACC responses. This apparent discrepancy may be explained by recent electrophysiological findings showing that contralateral preference can emerge in specific ACC neuronal subpopulations. For example, Koga et al. (2024) demonstrated that ACC neurons projecting to the contralateral claustrum preferentially respond to contralateral mechanical stimulation⁴⁸, suggesting that contralateral bias exists at the level of circuit-specific activity rather than global ACC function.

Importantly, most previous studies characterizing ACC function have been conducted under chronic pain conditions, where prolonged pathological states involve complex bilateral and affective processes^49,50. In contrast, our study examined responses at 1 day after drug administration, which likely reflects an acute or early-phase pain state. We therefore suggest that contralateral bias in the ACC may be more prominent during acute nociceptive processing, when sensory-driven activity is dominant, whereas bilateral and affective integration may become more pronounced during chronic stages. However, since our observations are limited to an early time point, we cannot exclude the possibility that ACC responses may shift toward a more bilateral pattern as pain persists over time.

It is important to note that the medial insular cortex neurons were largely affected in our ablation experiments, and the claustrum, located medially, has been recently implicated in pain processing⁵¹. Therefore, precise regional delineation was not fully achieved in the present study, which should be considered when interpreting the results.

Because the exact flow of information between the insular cortex and ACC remains unclear, future studies employing electrophysiological, optogenetic, or chemogenetic approaches will be necessary to fully elucidate the circuit mechanisms underlying LPA-induced MIP.

Moreover, our observations that microglia responses are associated with signalling through the corpus callosum and that minocycline suppresses this pain behaviour suggest that minocycline may merit further investigation as a therapeutic option for abnormal pain states, including MIP. Minocycline, a widely used tetracycline antibiotic, has a favourable safety profile⁵². Although some clinical studies support its analgesic potential, its efficacy remains inadequately established, and it is not yet widely used in pain management^53,54.

One possible explanation is the context-dependent nature of microglial function. For instance, in ischaemic stroke, microglia initially promote inflammation and neuronal injury but later aid in tissue repair and neuroprotection⁵⁵. This functional plasticity underscores the importance of timing when targeting microglial activation. Historically, microglia have been classified into M1 (pro-inflammatory) and M2 (anti-inflammatory) phenotypes⁵⁶. However, recent single-cell RNA sequencing has revealed substantial heterogeneity and functional diversity among microglial populations, including those in the corpus callosum^57,58. These microglia may support myelin repair and remyelination or modulate myelin homeostasis when repair mechanisms are compromised⁵⁸. In our study, inhibition of contralateral CD68/myeloid cell-positive cells improved MIP. Because AAV-CD68-hM4 has been reported to suppress pro-inflammatory cytokines, including IL-1β, IL-6, and TNFα²², the microglia/myeloid cell responses involved here may reflect a pro-inflammatory state. Our data are therefore consistent with maladaptive LPA-associated microglial activation, but the precise polarisation state remains to be determined. Thus, targeting LPA signalling may offer a promising strategy for modulating microglial function and alleviating abnormal pain.

In this study, we demonstrated that microglial responses contribute to LPA-associated amplification in the corpus callosum. However, this study has some limitations. First, we used only male mice. Given the known sex differences in microglial function and pain processing, the generalisability of the findings to females remains unclear. Future studies are required to determine whether the LPA–microglia pathway and the mirror-image pain phenotype are sexually dimorphic.

Additionally, this study reveals that callosal proximity is associated with propagated endogenous LPA production and bilateral hyperalgesia. However, because complete control experiments were not performed at every condition, caution is required to avoid overinterpreting small differences between nearby injection sites (Fig. 3D–K).

Moreover, we demonstrated analgesic effects of minocycline and PF-8380. However, because both drugs exert pleiotropic actions, the possibility remains that subtle non-nociceptive effects also influence pain responses.

Furthermore, this study suggests that callosal microglial activation contributes to MIP via the contralateral insula–ACC pathway. However, because minocycline was administered systemically, the precise microglial population and anatomical location responsible for the ACC effect remain unclear.

Finally, our present pharmacological data do not establish a microglia-specific LPA₁ pathway, and the detailed molecular mechanisms by which LPA-microglia interactions are linked to signals in the contralateral cortex remain unclear. In particular, the involvement of astrocytes in propagated LPA-related responses cannot be excluded. Future studies employing microglia-specific conditional knockout models will be essential to elucidate the molecular pathways involved.

In summary, our data indicate that LPA production associated with the corpus callosum contributes to MIP development and bilateral hyperalgesia. These findings provide mechanistic insight into central post-stroke pain and support further evaluation of LPA-related pathways and microglia responses as potential therapeutic targets.

Methods

Animals and ethical approval

A total of 234  Male C57BL/6J mice (6–7 weeks old, 20–22 g) were used in this study. The mice were obtained from Japan SLC, Inc. (Shizuoka, Japan) and housed under controlled environmental conditions: temperature of 24 ± 3 °C, humidity of 50 ± 10%, and a 12 h light/dark cycle (lights on: 08:00–20:00, lights off: 20:00–08:00). Food and water were provided ad libitum. All experimental procedures were approved by the Institutional Animal Care and Use Committees of Kyoto University (approval numbers: Med Kyo 24090, 25089) and Nagasaki University (approval number: 1607201325-8). All animal care and experimental protocols were conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines⁵⁹. Pain-related experiments adhered to the ethical recommendations of the International Association for the Study of Pain⁶⁰.

Drug administration

Rose Bengal (FUJIFILM Wako Chemicals, CAS No. 632-69-9) was dissolved in saline to a final concentration of 3 mg/mL and administered via tail vein injection at a dose of 30 mg/kg body weight. Oleoyl-lysophosphatidic acid sodium salt (18:1-LPA; Avanti Polar Lipids, Alabaster, AL, USA), prostaglandin E₂ (Nacalai Tesque, Kyoto, Japan), and oleoyl-lysophosphatidic acid-d₅ (18:1-LPA-d₅; Echelon Biosciences, UT, USA) were dissolved in aCSF (composition: 125 mM NaCl, 3.8 mM KCl, 1.2 mM KH₂PO₄, 26 mM NaHCO₃, 10 mM glucose, pH 7.4) to a final concentration of 2 mM. tPA (Activacin®, Kyowa Kirin, Tokyo, Japan) was reconstituted in the supplied sterile water for injection at approximately 1 mg/mL (equivalent to 6 million international units [IU]) and administered via tail vein injection at a dose of 10 mg/kg. PF8380 was dissolved in aCSF containing 25% polyethylene glycol, 20% 2-hydroxypropyl-β-cyclodextrin (Nacalai Tesque, Kyoto, Japan), and 2.5% dimethyl sulfoxide (DMSO, Nacalai Tesque), and administered intracerebroventricularly at a dose of 15 nmol in a 5 μL volume. AM966 was dissolved in DMSO to a final concentration of 10 mM and administered intracerebroventricularly at a dose of 1 nmol in a 5 μL volume. Minocycline was dissolved in PBS to a concentration of 4.5 mg/mL and administered intraperitoneally at a dose of 45 mg/kg body weight.

Ischaemia/reperfusion-induced CPSP model

The CPSP model was established using a previously reported PIT method^14,15. Briefly, mice were anesthetised with isoflurane, after which the fur on the left temporal region was shaved, and the skin incised to expose the underlying temporal muscle. The muscle was gently detached from the skull to allow visualisation of the middle cerebral artery (MCA) through the bone. The skull over the MCA was carefully thinned using a dental drill to expose the artery. Rose Bengal (30 mg/kg) was then administered via tail vein injection. Green light (5000 lux, wavelength 540 nm) was applied for 10 min to the exposed MCA using an L-4887 light source to induce thrombotic infarction. Although this infarction typically persists for more than 24 h ref. ¹⁴, reperfusion was achieved in this study by administering tPA via tail vein injection 6 h after infarction.

Stereotaxic microinjection

For intracerebral injections, mice were anesthetised with isoflurane using a Small Animal Anesthetizer (MK-AT210D, Muromachi Kikai Co., Ltd., Tokyo, Japan). Anaesthesia was induced with 4–5% isoflurane and maintained at 2% throughout the procedure. The head was secured in a stereotaxic frame (NARISHIGE, Tokyo, Japan), and the scalp was incised to expose the skull. The target brain region was located using a stereotaxic manipulator and appropriate coordinates. The skull overlying the target area was carefully drilled using the Ideal Micro-Drill™ (CellPoint Scientific Inc., MD, USA) to expose the brain surface. A glass capillary was inserted to the target depth from the brain surface, and reagents were delivered at a controlled rate (50–100 nL/min) using a UMP3 microsyringe pump controlled by a Micro-4 controller (World Precision Instruments, FL, USA).

Nociceptive test

Thermal stimulation test

Thermal nociception was assessed using the Hargreaves method, as previously described^14,61. In this method, a heat source is applied to the plantar surface of the hind paw, and the latency to paw withdrawal is recorded as a measure of thermal pain sensitivity. Prior to testing, the height of the glass platform and the intensity of the radiant heat source were calibrated such that untreated control mice exhibited a mean withdrawal latency of 10 ± 2 s. Mice were placed individually on the glass platform and confined within transparent plastic chambers (10 × 10 × 10 cm). They were allowed to habituate for at least 1 h to minimise stress and ensure accurate measurements. Mice that did not exhibit calm behaviour within 2 h were excluded from the experiment. Each hind paw was tested three times, with a minimum interval of 10 min between the trials to prevent sensitisation. The average of the three measurements was used as the final withdrawal latency. To avoid tissue damage, a maximum cutoff time of 20 s was imposed for each stimulus.

Mechanical paw pressure test

Mechanical nociception was evaluated using a digital von Frey apparatus (IITC), following previously reported procedures^14,62. A rigid probe with a 0.8 mm tip was applied perpendicularly to the plantar surface of the hind paw, and the pressure at which a withdrawal response occurred was recorded. Mice were placed on a metal mesh platform and housed individually within plastic chambers (10 × 10 × 10 cm) for at least 1 h to acclimate before testing. Each hind paw was tested three times, and the mean value was calculated as the final mechanical threshold. To minimise the risk of sensitisation or learned behaviour, measurements were spaced at least 10 min apart.

Accelerating Rotarod test

Motor coordination and balance were evaluated using the accelerating Rotarod test, based on our previously described method⁶³. Briefly, mice were placed on a Rotarod treadmill (Muromachi Kikai Co., Ltd., Tokyo, Japan), and the rotation speed was gradually increased to 45 rpm over 300 s. The maximum observation time was 5 min. Mice were trained for three consecutive days (12 trials in total) before PIT-induced cerebral ischaemia. During the test period (day 3 after PIT), mice underwent four trials per day. The latency to fall from the Rotarod was recorded, and the mean value was used for analysis.

Desorption electrospray ionisation-mass spectrometry imaging

Freshly dissected brain blocks were rapidly frozen using powdered dry ice and stored at -80°C until analysis. Tissue blocks were sectioned at −20 °C using a cryostat (CM1950, Leica Microsystems, Wetzlar, Germany) to obtain 8–10 μm-thick coronal sections. The prepared sections were mounted onto uncoated glass slides (Matsunami). Slides were analysed using the DESI XS system coupled to a Xevo TQ Absolute tandem quadrupole mass spectrometer (Waters, Milford, MA, USA). During imaging, charged microdroplets were sprayed onto the tissue surface to desorb and ionise molecules, which were then detected by the mass spectrometer. Spatial molecular information was obtained by mapping the positional coordinates of detected m/z values. The ions of interest included 18:1-LPA (precursor m/z 435.1, product ion m/z 153), 18:2-LPA (precursor m/z 433.1, product ion m/z 153), 18:0-LPA (precursor m/z 437.1, product ion m/z 153), 20:4-LPA (precursor m/z 457.1, product ion m/z 153), 18:1-LPA-d₅ (precursor m/z 440.1, product ion m/z 158), PGE₂ (precursor m/z 351.1, product ion m/z 235), and PGD₂ (precursor m/z 351.1, product ion m/z 233).

Immunohistochemistry

Mouse brains were rapidly frozen on powdered dry ice immediately after dissection. Coronal sections (10 µm thick) were prepared using a cryostat (Leica Microsystems, Wetzlar, Germany), mounted onto glass slides (Matsunami, Osaka, Japan), and stored at −80 °C until staining. Prior to staining, sections were air-dried for 1 h. Tissues were fixed in 4% paraformaldehyde (PFA) for 10 min and washed three times with phosphate-buffered saline (PBS) or PBS containing 0.1% Tween-20 (PBST) for 5 min each. Blocking was performed with 10% normal donkey serum for 30 min. Sections were incubated overnight at 4 °C with the following primary antibodies diluted as indicated: anti-Iba-1 (1:500; 019-19741, WAKO, Osaka, Japan), anti-CD68 (1:500; MCA1957GA, Bio-Rad, Hercules, CA, USA), anti-NeuN (1:1000; ab104224, Abcam, Tokyo, Japan), anti-GFAP, clone GA5 (1:500, MAB3402, Merck Millipore), anti-COX2 (1:500, Abcam, ab179800), and anti-CD11b, clone 5C6 (1:500, Bio-Rad). Following three additional washes with PBS or PBST, sections were incubated for 1 h at room temperature (24 ± 2 °C) with Alexa Fluor™-conjugated secondary antibodies (1:500; Thermo Fisher Scientific, Waltham, MA, USA). Finally, slides were rinsed and coverslipped for imaging. For c-Fos staining, mouse brains were fixed by perfusion with 4% PFA and cryoprotected in 30% sucrose. Coronal sections (30 µm thick) were prepared using a cryostat (Leica Microsystems, Wetzlar, Germany) in PBS. Blocking was performed with 10% normal donkey serum for 1 h. Sections were incubated for 3 h at room temperature with anti-c-Fos (1:500; ab190289, Abcam). Following three additional washes with PBST, sections were incubated for 30 min at room temperature with Alexa Fluor™-conjugated secondary antibodies (1:1000; Thermo Fisher Scientific). Finally, slides were coverslipped for imaging.

Nissl staining

Fresh-frozen tissue sections (10 μm) were fixed in 4% paraformaldehyde (PFA) and subsequently washed. The sections were then immersed in Nissl staining solution　(Cresyl violet acetate, 10510-54-0, MP Biomedicals, LLC, CA, USA) and incubated at 37 °C for 10 min. After staining, differentiation was performed in 95% ethanol containing a few drops of 10% acetic acid for 10 min, followed by further differentiation in 95% ethanol and 100% ethanol for 1 min each. The sections were subsequently dehydrated in 100% ethanol two times and cleared in xylene three times.

Fluorescence image acquisition and cell counting

Fluorescence images were acquired using an all-in-one fluorescence microscope (BZ-X, KEYENCE, Osaka, Japan) and a FLUOVIEW FV3000 confocal microscope (Olympus, Tokyo, Japan). For c-Fos counting, high-magnification images (×40) acquired from the white boxed regions were analysed by manual counting of positive cells using ImageJ software.

Statistical analyses

All quantitative data are expressed as mean ± standard error of the mean. Statistical analyses were performed using GraphPad Prism version 8.0 (GraphPad Software, San Diego, CA, USA). Comparisons between two groups were conducted using unpaired two-tailed Student’s t-test. For comparisons involving more than two groups, two-way analysis of variance followed by Bonferroni’s post hoc multiple comparison test was used. A P value of less than 0.05 was considered statistically significant.

Author contributions

Conceptualisation: H.U. and Y.S. Methodology: H.U. and Y.S. Software: Y.S. Validation: H.N., R.K., and R.M. Formal analysis: H.N. Investigation: H.N., R.K., and R.M. Resources: Y.S. Data curation: H.N. and Y.S. Writing-original draft: H.N., H.U., and Y.S. Writing-review and editing: H.U. and Y.S. Visualisation: H.N. and Y.S. Supervision: H.U. and Y.S. Project administration: H.U. and Y.S. Funding acquisition: H.N., H.U., and Y.S.

Peer review

Peer review information

Communications Biology thanks Hyoung-Gon Ko and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Alban Latremoliere and Benjamin Bessieres. A peer review file is available.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information. Source data underlying the graphs in the main figures are provided in Supplementary Data.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10261-5.