CREB–KIF1A–CGRP-positive feedback loop drives central sensitization in chronic migraine

Abstract

Background

Chronic migraine (CM), defined as ≥ 15 monthly headache days and often involving central sensitization, is a complex neurological disorder. While calcitonin gene-related peptide (CGRP) is a key migraine target, its upstream regulation remains unclear. Kinesin family member 1A (KIF1A), which is essential for axonal transport, is transcriptionally regulated by cAMP response element-binding protein (CREB). The aim of this study was to elucidate the role of the CREB–KIF1A–CGRP signaling axis in the pathogenesis of migraine.

Methods

CMs were generated in mice through repeated nitroglycerin (NTG) injections, and thermal and mechanical allodynia was evaluated behaviorally. Molecular changes in the spinal trigeminal nucleus caudalis (SP5C) and Neuro-2a cells were examined using immunoblotting, qPCR, and immunofluorescence. Mechanistic studies included chromatin immunoprecipitation and dual-luciferase assays to verify the CREB-mediated transcriptional regulation of Kif1a, coimmunoprecipitation (Co-IP) to assess KIF1A–CGRP interactions, and synaptosomal CGRP analysis following Kif1a knockdown. Pharmacological interventions included Forskolin (a CREB agonist), 666 − 15 (a CREB inhibitor), Kif1a knockdown/overexpression, and Olcegepant (a CGRP receptor antagonist).

Results

NTG treatment activated the CREB–KIF1A–CGRP pathway and induced migraine-like hypersensitivity. Forskolin-induced CREB activation upregulated KIF1A and CGRP expression, whereas inhibition of CREB expression by 666 − 15 reversed these effects. ChIP and luciferase assays confirmed the direct binding of CREB to the Kif1a promoter. Kif1a knockdown reduced p-CREB, CGRP, and c-Fos levels in SP5C tissue and Neuro-2a cells, alleviating behavioral hypersensitivity. Co-IP showed that KIF1A physically associated with CGRP, and synaptosome analysis revealed decreased CGRP in vesicles after Kif1a knockdown. Overexpression of Kif1a increased CGRP levels, but this effect was reversed by age. CGRP receptor blockade also inhibited Forskolin-induced activation of p-CREB, c-Fos, and KIF1A. These findings support a CGRP-dependent positive feedback loop within the CREB–KIF1A–CGRP axis that drives migraine pathogenesis.

Conclusions

These findings reveal a CREB–KIF1A–CGRP positive feedback loop that drives migraine-like hypersensitivity. CREB directly regulates KIF1A transcription, which in turn promotes CGRP expression and signaling. Disruption of this axis effectively attenuated migraine-associated behaviors and molecular responses, highlighting potential therapeutic targets for the treatment of CM.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s10194-025-02147-4.

Background

Migraine is a highly disabling neurological disorder that affects more than one billion individuals worldwide and is among the leading causes of years lived with disability, particularly among young and middle-aged adults [1]. Chronic migraine (CM), a highly disabling subtype of migraine, can cause patients to be unable to work and live normally, severely affecting their quality of life [2, 3]. Newly emerging evidence suggests that central sensitization is related to the pathophysiological mechanism of chronic migraine [4]. The mechanisms of central sensitization involve alterations in neuronal excitability, ion channel function, and the release of neurotransmitters and neuropeptides [5]. These alterations occur across multiple regions of the central nervous system, including the spinal trigeminal nucleus caudalis (SP5C), a key site for processing cephalofacial nociceptive inputs. However, the molecular and cellular mechanisms underlying migraine chronification remain poorly understood, and the current treatment options are suboptimal for many patients.

A key breakthrough in migraine therapeutics has been the identification of calcitonin gene-related peptide (CGRP) as a central molecular mediator [6]. CGRP is a neuropeptide that is abundantly expressed in the trigeminovascular system, where it facilitates neurogenic inflammation, peripheral sensitization, and central pain transmission. Monoclonal antibodies and small-molecule antagonists targeting CGRP or its receptor have shown clinical efficacy in both acute and preventive treatment of migraine [7]. Nevertheless, the response to CGRP-targeted therapy is highly variable, and a considerable proportion of patients either fail to respond or experience relapse after the cessation of treatment [8, 9]. These findings suggest that CGRP-related pathways are modulated by additional upstream or parallel mechanisms that contribute to the persistence and amplification of migraine signaling.

The kinesin superfamily comprises a broad group of neuron-specific motor proteins. KIF1A is a major axonal transport motor protein involved in the selection and regulation of dense core vesicle, lysosomal and synaptic vesicle trafficking [10–12]. cAMP response element-binding protein (CREB) activation leads to the transcription of numerous genes implicated in activity-dependent neuronal plasticity, long-term potentiation, and nociceptive sensitization [13]. Our previous research indicated that activated CREB upregulates KIF1A, leading to neuropathic pain [14]. CGRP processing involves its trafficking from the endoplasmic reticulum to dense core secretory vesicles. Secretory vesicles are then transported downward along the axon [6]. Emerging studies have highlighted the roles of transcriptional regulation and axonal transport in the modulation of pain-related signaling cascades. KIF1A plays a crucial role in maintaining synaptic transmission and neuropeptide release by ensuring the efficient delivery of cargo, including CGRP, to peripheral and central terminals.

Given the associations among CREB, KIF1A, and CGRP, we hypothesized that KIF1A facilitates CGRP trafficking and release and that CGRP may, in turn, increase CREB activation through receptor-mediated signaling, resulting in the formation of a self-sustaining positive feedback loop. This loop can continuously amplify CGRP signaling within the trigeminovascular system, contributing to the maintenance of sensitization and the transition to CM. We employed a well-established nitroglycerin (NTG)-induced chronic migraine mouse model, which replicates key behavioral and molecular features of human migraine, to investigate this hypothesis. We assessed the dynamics of CREB, KIF1A, and CGRP expression in the SP5C—a critical relay station for craniofacial nociceptive processing—and evaluated the effects of pharmacological modulators targeting each component of the proposed pathway. Furthermore, we examined the behavioral consequences of manipulating this feedback loop in vivo. Our findings reveal a previously unrecognized CREB–KIF1A–CGRP regulatory circuit that contributes to migraine chronification and identify KIF1A as a potential upstream therapeutic target beyond CGRP blockade.

Methods

Animal preparation

A total of 151 male C57BL/6 mice (143 male mice and 8 female mice), aged 7 to 8 weeks and weighing between 20 and 25 g, were utilized in this study. We primarily used male mice in our experiments, except for the experiment involving the initial validation of CREB–KIF1A-CGRP circuit activation, which included female mice. The use of males was intended to minimize hormonal fluctuations associated with the estrous cycle, thereby reducing biological variability and enhancing data consistency. This strategy improved the interpretability of migraine-related findings by avoiding the additional complexity introduced by sex-specific neuroendocrine factors. Moreover, male mice are known to exhibit more consistent behavioral responses in nociceptive and stress-related models, facilitating more accurate assessments of migraine-like phenotypes and pharmacological interventions.

All the animals were sourced from Vital River Laboratory Animal Technology (Beijing, China) and maintained in a specific pathogen-free environment at the Animal Experimental Center of the First Hospital of Jilin University. The mice were housed under standard laboratory conditions, with the ambient temperature controlled at 22 ± 2 °C and the relative humidity maintained at 60 ± 5%, on a 12-hour light/dark cycle. Food and water were provided ad libitum throughout the experiment. All animal handling and experimental procedures complied with the institutional and national ethical guidelines for the care and use of laboratory animals. The study protocol received prior approval from the Institutional Animal Care and Use Committee of Jilin University, and all efforts were made to reduce animal suffering and minimize the number of animals used.

Establishment of the migraine model

NTG is a well-established pharmacological tool for modeling migraine in both clinical and preclinical settings [15]. Systemic administration of NTG elicits migraine-like manifestations, such as enhanced nociceptive sensitivity and behavioral responses linked to specific brain areas, including the striatum [16]. In this study, NTG was utilized to induce a CM-like state for mechanistic investigations. A commercial NTG stock solution (5 mg/mL; Beijing Yimin Pharmaceutical Co., Ltd., Beijing, China) was diluted with 0.9% saline to yield a final concentration of 1 mg/mL. The mice received intraperitoneal (i.p.) injections of NTG at a dosage of 10 mg/kg on alternate days.

Behavioral testing methods

The hindpaw mechanical threshold (HMT) and periorbital mechanical threshold (PMT) were evaluated using von Frey filaments, and the up-down method was used to calculate the 50% withdrawal threshold [17]. The mice were positioned on an elevated wire mesh platform, and calibrated filaments were applied sequentially to the hindpaw and periorbital regions to elicit withdrawal responses indicative of mechanical hypersensitivity. Testing commenced with a 0.008 g filament for the periorbital area and a 0.6 g filament for the hindpaw, and the filament force was gradually increased until a positive response—brisk withdrawal—was observed [17, 18]. Each animal underwent at least three rounds of testing, with 1 to − 2 min between stimuli to prevent sensitization. All the assessments were performed by experimenters who were blinded to the group assignments. Hindpaw thermal latency (HML) was measured using a BME-410 C automatic thermal pain stimulator (Chinese Academy of Medical Science and Peking Union Medical College Institute of Biomedical Engineering, Tianjin, China). A cutoff time of 20 s was set to prevent tissue damage to the hindpaws. Each mouse was tested five times on the designated side, with 5-minute intervals between tests to ensure consistent and reliable measurements.

Animal experimental groups

The animal experiment was divided into four parts. Specifically, in Experiment 1, the mice were divided into two groups: the (1) Vehicle (VEH) and (2) NTG groups. In Experiment 2, the mice were divided into six groups: the (1) VEH, (2) NTG, (3) Wild-type (WT) + Forskolin (FSK) (1 mg/kg) (a specific CREB agonist), (4) WT + FSK (2 mg/kg), (5) WT + FSK (5 mg/kg) and (6) WT + DMSO groups. The groups in Experiment 3 were the (1) VEH + DMSO, (2) VEH + 666 − 15 (a specific CREB), (3) NTG + DMSO, and (4) NTG + 666 − 15 groups. In Experiment 4, the groups were as follows: the (1) VEH + MS siRNA, (2) VEH + Kif1a siRNA, (3) NTG + MS siRNA, and (4) NTG + Kif1a siRNA groups. These groups are illustrated in Fig. 1.

Fig. 1.

Experimental design and timeline for behavioral testing and tissue collection. A VEH and NTG groups: The mice received NTG or saline every other day (i.p.) on days 1, 3, 5, 7, and 9. Behavioral testing was conducted prior to each injection, and tissue collection was performed on day 10. B FSK and NTG groups: Mice were intraperitoneally injected with either FSK (1, 2, or 5 mg/kg) or NTG (10 mg/kg) on days 1, 3, 5, 7, and 9. The VEH group for NTG received an equal volume of saline, whereas the control group for FSK received an equivalent volume of DMSO. Behavioral testing and tissue collection were conducted as described in (A). C 666–15 treatment protocol: On days 1, 3, 5, 7, and 9, the mice were first administered NTG (10 mg/kg, i.p.) or vehicle (saline), followed by an injection of 666–15 (10 mg/kg, i.p.) or an equivalent volume of DMSO as the control. Behavioral testing and tissue collection were performed as described in (A). D Kif1a siRNA group: The mice were intrathecally injected with Kif1a siRNA or a mismatch control siRNA daily from days 1 to 9, concurrent with NTG or saline injections. Behavioral testing was performed on alternate days, and tissue collection was performed on day 10

Drug administration to animals

Based on previous studies, the selected doses of the FSK (MedChemExpress, USA) were 1, 2, and 5 mg/kg [19]. WT mice received intraperitoneal (i.p.) injections of either FSK, diluted in 5% DMSO to the desired concentration, or vehicle control (5% DMSO alone) every other day. VEH-treated mice were injected at the same volume and time points those in the FSK group to ensure consistent treatment conditions. Similarly, the selected dose of the CREB inhibitor 666 − 15 (MedChemExpress, USA) was 10 mg/kg according to established protocols [20, 21]. 666 − 15 was dissolved in dimethyl sulfoxide (DMSO), and 200 µL of the solution or 5% DMSO was administered intraperitoneally 30 min prior to each NTG injection. The mice in the VEH group received 5% DMSO alone in an identical manner.

Kif1a-specific small interfering RNA (siRNA) was designed and synthesized by Genecfps (Suzhou, China). A nontargeting mismatch siRNA (MS-siRNA) with no sequence homology to any known mouse gene was used as a negative control. Both siRNAs were administered under identical conditions (dose, volume, and injection schedule) to ensure comparability between groups. Kif1a-siRNA and MS-siRNA were diluted to a concentration of 5 µg/10 µl per mouse in diethyl pyrocarbonate-treated water. The Kif1a siRNA sequence was validated in vitro, as described in detail in the Cell Culture section. The administered siRNA concentration was selected based on our previously published protocols [14].

For simulated preventive treatment, Kif1a-siRNA or MS-siRNA was administered via intrathecal injection 30 min prior to each NTG injection. The specific siRNA sequences are provided in Table 1. Details of the drug administration protocol are illustrated in Fig. 1. The mice were anesthetized with isoflurane, and intrathecal injections were performed using a Hamilton microsyringe inserted into the L5–L6 intervertebral space. Following injection, the syringe was held in place for 10 s to facilitate proper diffusion. Successful intrathecal delivery was confirmed by a characteristic tail‒flick reflex or mild lateral body movement. The injection volume was limited to 10 µL to minimize the risk of systemic exposure. No motor impairments or off-target effects were observed following the procedure.

Table 1.

Nucleotide sequences used in this study

Names	Sequences (5′ to 3′)
Real-time PCR
Kif1a F	GGA CAC ATC AGT AGC CTA TGT CC
Kif1a R	ACC TCC TGT AAG AGG CTC AGT TT
ChIP–qPCR
Kif1a F	TGC TGC ACC CGA TTA CGG
Kif1a R	CAA AGC GGC ACA GAA GAG AC
Kif1a-1 siRNA Kif1a-1
Sense	CAA CAA GCU GAU UAG AGA TT
Antisense	UUC UCU AAU CAG CUU GUU GTT
Kif1a-2 siRNA Kif1a-2
Sense	GGA GCU GCG CAU UGU AGA ATT
Antisense	UUC UAC AAU GCG CAG CUC CTT
Kif1a-3 siRNA 3
Sense	CCA UGC UCA AGA CAC CCA ATT
Antisense	UUG GGU GUC UUG AGC AUG GTT

F forward, R reverse, PCR polymerase chain reaction

Neuro-2a neuroblastoma cell culture and treatments

Neuro-2a (N2A) cells, kindly provided by the Neurology Laboratory of the First Hospital of Jilin University and originally obtained from ATCC (CBP60307), were cultured in DMEM supplemented with 10% FBS, 100 U/mL penicillin, and 100 U/mL streptomycin, and maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

Cultured cells were stimulated with the adenylate cyclase activator FSK at a concentration of 10 µM to investigate cAMP-dependent signaling dynamics [22]. FSK was initially dissolved in DMSO and stored at − 20 °C. The cells were exposed to FSK for four different durations, namely, 2 h, 6 h, 12 h, and 24 h. A single vehicle control group (0.1% DMSO) was included and processed in parallel under the same experimental conditions.

Cells were treated with the selective CGRP receptor antagonist Olcegepant (MedChemExpress, USA) to assess the role of CGRP receptor signaling in the effects of KIF1A. Olcegepant was first dissolved in DMSO to generate a stock solution and then diluted in culture medium to final working concentrations of 10, 20, 50, and 100 nM. A VEH group received an equivalent volume of DMSO diluted in medium to control for solvent-related effects. The cells were incubated with Olcegepant for 2 h under standard culture conditions (37 °C, 5% CO₂) prior to WB analyses to assess changes in CGRP and KIF1A expression.

Three siRNA sequences targeting Kif1a were designed and synthesized by Genecfps (Suzhou, China), along with a mismatch siRNA (MS-siRNA) that served as a negative control. Cells were transfected in 6-well plates with 5 µL of Kif1a siRNA per well to achieve gene knockdown. The most effective siRNA sequence was selected for in vivo experiments by evaluating the knockdown efficiency by quantitative reverse transcription polymerase chain reaction (qRT‒PCR) and Western blot (WB) analyses.

A Kif1a expression plasmid was designed and synthesized by Genecfps (Suzhou, China) to overexpress Kif1a. The coding sequence of mouse Kif1a was cloned and inserted into the pcDNA3.1(+) vector. For transfection, cells were seeded in 6-well plates and transfected with either the Kif1a-pcDNA3.1 overexpression plasmid or an empty pcDNA3.1 vector, which served as the negative control. Transfections were performed using Gencefe TransExp reagent (Genecfps, China) according to the manufacturer’s instructions. After 24–48 h, the cells were harvested for downstream analyses, including qPCR and WB, to verify the overexpression efficiency.

WB analysis

The mice were humanely euthanized using an overdose of carbon dioxide. The SP5C was carefully dissected, immediately frozen in liquid nitrogen, and stored at − 80 °C until subsequent analyses. Animal tissues were lysed on ice in lysis buffer. The cells were subsequently washed twice with cold phosphate-buffered saline (PBS) and lysed on ice with lysis buffer. The protein concentrations in the extracted lysates were determined using the bicinchoninic acid assay (BCA). Total protein (20–40 µg) was subsequently separated by SDS‒PAGE and transferred onto polyvinylidene fluoride membranes. The membranes were blocked with 5% nonfat dry milk for 2 h at room temperature and then incubated overnight at 4 °C with the appropriate primary antibodies. The primary antibodies used include rabbit anti-c-Fos (Abcam, United Kingdom); rabbit anti-CGRP (CST, USA); rabbit anti-CREB (Abcam, United Kingdom); rabbit anti-p-CREB (S133) (Abcam, United Kingdom); mouse anti-KIF1A (BD Biosciences, USA); rabbit anti-Synaptophysin (Abcam, United Kingdom); rabbit anti-PSD-95 (Cohesion, China); rabbit anti-β-actin (Cohesion, China); and rabbit anti-GAPDH (Cohesion, China). After being washed, the membranes were incubated for 1 h with anti-mouse or anti-rabbit secondary antibodies (Proteintech, China). The membranes were analyzed using an imaging system. Band intensities were quantified using ImageJ software (NIH, USA). Signal intensities were normalized to the corresponding loading control (GAPDH), and the results are expressed as relative density. For WB analysis, samples prepared from six biologically independent mice per group (n = 6) were used.

Immunofluorescence (IF) staining of frozen sections

The mice were deeply anesthetized and transcardially perfused with PBS. SP5C tissues were then collected, fixed overnight with 4% paraformaldehyde at 4 °C, and subsequently cryoprotected in a 30% sucrose solution until they sank. The samples were embedded in OCT compound and sliced into 10 μm sections using a cryostat. The sections were blocked with 10% normal goat serum and incubated overnight at 4 °C with primary antibodies, including rabbit anti-c-Fos (Abcam, United Kingdom), rabbit anti-KIF1A (Proteintech, China), rabbit anti-CGRP (Abcam, United Kingdom), and mouse anti-NeuN, anti-Iba1, and anti-GFAP (NeuN and GFAP from Cell Signaling Technology, USA; Iba1 from Santa Cruz, USA) antibodies. Following thorough washes with PBS, the sections were treated with fluorophore-labeled secondary antibodies for 40 min at room temperature. Nuclei were counterstained accordingly. IF staining was performed on cryosections obtained from four biologically independent mice per group (n = 4).

IF staining of cells

Neuro-2a cells were seeded on glass coverslips and cultured to 50% confluence. After treatment, the cells were washed three times with PBS, followed by fixation with 4% paraformaldehyde. After permeabilization with 0.2% Triton X-100/PBS, nonspecific binding was blocked with 5% BSA. Primary antibodies ((rabbit anti-KIF1A (Proteintech, China) and rabbit anti-CGRP (Abcam, United Kingdom)) diluted in blocking buffer were applied overnight at 4 °C. The subsequent procedures were identical to those described for frozen tissue sections. Fluorescence images were captured using a confocal microscope.

IF staining of paraffin-embedded sections

Multiplex IF staining was performed on FFPE tissue sections using a TSA system (Servicebio, China). The sections were deparaffinized, rehydrated, and subjected to antigen retrieval by heating in a microwave. Endogenous peroxidase activity was blocked with 3% H₂O₂, followed by blocking with serum or BSA. Primary antibodies (rabbit anti-KIF1A (Proteintech, China) and rabbit anti-p-CREB (Abcam, United Kingdom)) were applied sequentially, and the samples were incubated overnight at 4 °C. HRP-conjugated secondary antibodies were added, and signal amplification was performed using iF488 or iF555 TSA. Between rounds, the antibodies were removed via heat-induced retrieval. The nuclei were counterstained with DAPI, and the sections were mounted with antifade medium. Fluorescence images were captured using a confocal microscope.

RNA isolation and real-time quantitative PCR

Total RNA was isolated from the collected cells using TransZol reagent (TransGen, China) according to the manufacturer’s instructions. Complementary DNA (cDNA) synthesis was performed with a Monad Reverse Transcription Kit (Monad Biotech, China). Quantitative real-time PCR (qRT‒PCR) was conducted using gene-specific primers (listed in Table 1). Each 20 µL reaction mixture consisted of 10 µL of 2× Color SYBR Green qPCR Master Mix (EZBioscience, China), 10 nM each of forward and reverse primers, and 50 ng of cDNA template. Amplification was performed according to the instructions of the qPCR kit. Gapdh served as the internal control, and relative mRNA expression levels were analyzed using the comparative Ct (2^^–ΔΔCt) method. qRT‒PCR was conducted on RNA samples isolated from six biologically independent mice per group (n = 6).

Chromatin immunoprecipitation–quantitative polymerase chain reaction

Chromatin immunoprecipitation (ChIP) was performed using the Pierce™ Magnetic ChIP Kit (Thermo Fisher Scientific). Mouse SP5C tissues were dissected and minced with ophthalmic scissors. The tissues were crosslinked with 1.5% formaldehyde for 10 min at room temperature, followed by quenching with 1× glycine. After centrifugation, the DNA was cleaved with lysis buffer containing Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific), followed by MNase cleavage into fragments averaging 200–1000 bp. The nuclear membranes were subsequently disrupted by sonication. For immunoprecipitation, 8 µL of rabbit anti-p-CREB antibody (Abcam, UK) or 1 µg of normal rabbit IgG control was added, and the mixture was incubated overnight at 4 °C with rotation. Input controls (10% of the sample) were reserved prior to IP. The purified DNA fragments were analyzed by qPCR using SYBR Green detection, and the primers used are listed in Table 1. ChIP followed by qPCR was performed using chromatin samples isolated from three biologically independent mice per group (n = 3).

Dual-luciferase reporter assay

The assay was performed by Genecfps (Suzhou, China). HEK293T cells (70–80% confluent in DMEM/10% FBS) were seeded into 12-well plates. Transfection complexes were prepared by mixing 500 ng pcDNA3.1 (empty) or pcDNA3.1-CREB1, 200 ng pGL4-Basic or pGL4-KIF1A-Promoter, and 20 ng pRL-TK (Renilla) in 100 µL serum-free DMEM. Separately, 4 µL Lipofectamine 2000 was mixed with 100 µL serum-free DMEM, incubated 5 min, then combined with the DNA mixture and incubated further 20 min. After 6 h, medium was replaced with fresh complete DMEM. Cells were lysed 48 h post-transfection using 250 µL 1× Passive Lysis Buffer/well. Firefly and Renilla luciferase activities were sequentially measured (Dual-Luciferase^® Reporter Assay System, Promega) on a SpectraMax iD3 plate reader. Firefly luciferase activity was normalized to Renilla luciferase activity, and relative promoter activity was expressed as fold change relative to the empty vector control. Assays were performed in triplicate and repeated independently three times (n = 3).

Coimmunoprecipitation assay

IP was performed using a Co-IP Kit (Absin, China). SP5C from three NTG mice was minced in lysis buffer, the supernatants were collected, and the samples were then incubated with 100 µL of anti-mouse CGRP (Santa Cruz, USA) on a rotator at 4 °C overnight. The following day, the antigen/antibody mixture was incubated for 2 h at room temperature with washed Protein A/G magnetic beads. Protein sample buffer was added after the magnetic beads had been separated from the bead-bound protein, and the mixture was boiled for 5 min. KIF1A was detected by WB with a rabbit anti-KIF1A antibody (Proteintech, China).

Isolation of synaptosomes from brain tissue

Synaptosomes were isolated from fresh or frozen brain tissue using the Minute™ Synaptosome Isolation Kit (Invent Biotechnologies, USA) with minor modifications. Approximately 30 mg of brain tissue was homogenized in ice-cold Buffer A containing protease and phosphatase inhibitors, followed by differential centrifugation steps at 4 °C to enrich the synaptosomal fractions. The final pellet was resuspended in 50 of buffer for downstream applications. Protein concentrations were determined using a BCA protein assay037, and equal amounts of synaptosomal protein were subsequently analyzed by WB. For synaptosomes WB analysis, samples prepared from three biologically independent mice per group (n = 3) were used.

Statistical analysis

Animals were randomly assigned to different experimental groups using a computer-generated sequence in Microsoft Excel to ensure objective group allocation. The sample size was determined in accordance with the recommendations of the ethics committee, balancing the need to reduce the number of animals used with the requirement for statistically meaningful outcomes. The data were processed using GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA). All the results are presented as the means ± standard errors (means ± SEMs). For comparisons between two groups, unpaired t tests were performed. When three or more groups were analyzed, one-way or two-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was applied. A p value of less than 0.05 was considered to indicate statistical significance. Sample sizes (n) for each experiment are indicated in the figure legends and were determined based on prior studies and standard practices in the field. In general, n refers to the number of animals per group (for in vivo experiments) or the number of independent replicates (for in vitro assays).

Results

Repeated injections of NTG activated the CREB–KIF1A–CGRP loop in the SP5C

A CM model was established using male and female mice by i.p. injection of NTG every other day for 9 consecutive days. The mechanical and thermal nociceptive thresholds of the hind paws, as well as the periorbital mechanical thresholds, were assessed both before each NTG or VEH injection (baseline) and 2 h after each injection (posttreatment response). Compared with VEH-treated controls, NTG-treated mice presented progressive reductions in hind paw mechanical and thermal thresholds, as well as in periorbital mechanical thresholds, starting on day 3, with the lowest levels being reached by day 9 (Fig. 2A–C for males; Supplementary Fig. 1A–C for females). WB analyses revealed significantly increased levels of p-CREB and KIF1A proteins in the SP5C of NTG-treated mice, whereas total CREB (t-CREB) levels remained unchanged (Fig. 2D–G for males; Supplementary Fig. 1D–G for females). Similarly, the protein expression of the c-Fos and CGRP proteins was markedly increased (Fig. 2D, F; Supplementary Fig. 1D, F). IF staining further revealed an increased density of CGRP-immunoreactive fibers (Fig. 2H–I) and increased c-Fos activity in the SP5C of NTG-treated male mice compared with VEH controls (Fig. 2J–K). In both male and female mice with CM, the KIF1A protein was activated, suggesting central sensitization following NTG administration. Subsequent animal experiments were conducted exclusively in male mice.

Fig. 2.

NTG upregulates CGRP, c-Fos, KIF1A, and p-CREB expression in the trigeminal system in male mice. A–C Compared with the VEH (saline) control (n = 8 mice per group), NTG induced mechanical allodynia (HML and HMT) and thermal hyperalgesia (PMT) compared with the VEH(Saline) control (n = 8 mice per group). D, E Representative WB images showing increased levels of KIF1A, c-Fos, p-CREB, t-CREB, and CGRP in NTG-treated mice. GAPDH was used as a loading control. F Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to GAPDH. G Ratio of p-CREB to t-CREB. (n = 6 mice per group). H, J Representative images of IF staining images showing CGRP (H) and c-Fos (J) expression in the SP5C. Nuclei were counterstained with DAPI (blue). Scale bars: 200 μm (H), 100 μm (J). I, K Quantification of the fluorescence intensity in (H) and (J), showing the significant upregulation of CGRP (I) and c-Fos (K) in NTG-treated mice (n = 4 mice per group). Statistical analysis: Two-way ANOVA for (A–C); unpaired two-tailed t test for (F, G, I, K). *p < 0.05, **p < 0.01, **p < 0.001 vs. the VEH group. All the data are presented as the means ± SEMs

CREB activation induces migraine-like behaviors in WT mice

We investigated the role of CREB in NTG-induced migraine in male mice. FSK, a known inducer of intracellular cAMP production, can activate CREB. We administered FSK intraperitoneally to WT mice to determine whether CREB activation could mimic migraine-like symptoms. Based on previous studies [19], we selected three doses—low, medium, and high (1, 2, and 5 mg/kg, respectively)—and administered the drug every other day for a total of five doses. The mechanical and thermal nociceptive thresholds of the hind paw, as well as the periorbital mechanical threshold, were measured 2 h after each injection.

The results revealed that the low dose induced a slight reduction in pain thresholds, but this change was not statistically significant (Fig. 3A–C). In contrast, both the medium and high doses resulted in significant decreases in the hind paw mechanical and thermal thresholds, as well as periorbital thresholds, starting after the second injection, with the effects persisting through day 9 (Fig. 3A–C). These findings indicate that medium and high doses of FSK can induce migraine-like symptoms. A dose of 2 mg/kg was selected for subsequent experiments to minimize potential toxicity.

Fig. 3.

CREB activation contributes to NTG-induced hyperalgesia and the upregulation of CGRP, c-Fos, and KIF1A expression. A–C FSK (1, 2, or 5 mg/kg, i.p.) and NTG induced mechanical allodynia (HML and HMT) and thermal hyperalgesia (PMT) (n = 8 mice). D–E Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (D), and t-CREB (E) in SP5C tissues. F Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to those of GAPDH. (G) Ratio of p-CREB to t-CREB. (n = 6 mice per group). H, J IF staining for CGRP (H) and c-Fos (J) in the SP5C. Nuclei were labeled with DAPI (blue); scale bars: 200 μm (H) and 100 μm (J). I,K Quantification of the fluorescence intensity of CGRP (I) and c-Fos (K) in each group (n = 4 mice per group). Statistical analysis: Two-way ANOVA for (A–C); one-way ANOVA for (F, G, I, K). *p < 0.05, **p < 0.01, and ***p < 0.001 compared with the VEH group. All the data are presented as the means ± SEMs

WB analyses showed that FSK significantly increased the levels of p-CREB (Fig. 3D, G), while t-CREB protein levels remained unchanged (Fig. 3E, G). In addition, FSK markedly upregulated the expression of the KIF1A, CGRP, and c-Fos proteins (Fig. 3D, F). IF staining further confirmed a significant increase in CGRP (Fig. 3H, I) and c-Fos (Fig. 3J, K) expression in the SP5C following the i.p. injection of the CREB agonist compared to DMSO-treated controls.

Inhibition of CREB inhibits KIF1A and CGRP expression in the SP5C

CREB expression was suppressed through i.p. injection of the CREB inhibitor 666 − 15 to investigate the role of the CREB–KIF1A–CGRP pathway in NTG-induced nociceptive hypersensitivity. This treatment significantly attenuated mechanical and thermal hypersensitivity, as well as periorbital pain, in male mice. In contrast, mice receiving i.p. injections of DMSO exhibited no significant changes in pain thresholds (Fig. 4A–C).

Fig. 4.

CREB inhibition attenuated NTG-induced hyperalgesia and ameliorated the upregulation of CGRP, c-Fos, and KIF1A expression. A–C NTG induced mechanical allodynia (decreased HML and HMT) and thermal hyperalgesia (decreased PMT). Compared with NTG + DMSO, 666 − 15 reduced nociception (n = 8 mice per group). D–E Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (D), and t-CREB (E) in SP5C tissues. F Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to those of GAPDH. (G) Ratio of p-CREB to t-CREB. (n = 6 mice per group). H, J IF staining for CGRP (H) and c-Fos (J) in the SP5C. DAPI (blue) was used to label the nuclei; scale bars: 200 μm (H) and 100 μm (J). I, K Quantification of the fluorescence intensity of CGRP (I) and c-Fos (K) across groups (n = 4 mice per group). Statistical analysis: Two-way ANOVA for (A–C); one-way ANOVA for (F, G, I, K). *p < 0.05, **p < 0.01, and ***p < 0.001 compared with the VEH + DMSO group; ^#p < 0.05, ^##p < 0.01, and ^###p < 0.001 compared with the NTG + DMSO group. All the data are presented as the means ± SEMs

WB analyses showed that 666 − 15 significantly reduced the levels of p-CREB (Fig. 4D, G), while t-CREB protein levels remained unchanged (Fig. 4E, G). The NTG-induced upregulation of KIF1A and CGRP was markedly inhibited by i.p. injection of the CREB inhibitor 666 − 15 in male mice (Fig. 4D, F). Similarly, the increase in c-Fos expression was also suppressed following 666 − 15 treatment (Fig. 4D, F). IF staining further confirmed that treatment with 666 − 15 significantly attenuated the expression of CGRP (Fig. 4H, I) and c-Fos (Fig. 4J, K) in the SP5C compared to DMSO-treated controls.

Kif1a Silencing suppresses CGRP and KIF1A expression in cultured N2A cells

The sequences of Kif1a siRNA were provided by Genecfps. We first identified the most efficient inhibitory sequence using the N2A cell line. Based on the WB (Fig. 5A) and qPCR (Fig. 5B) results, sequence 2 showed the highest knockdown efficiency. Therefore, sequence 2 was used for all subsequent cell and animal experiments. Compared with the Control group and the control + MS siRNA group, Kif1a knockdown not only reduced CGRP expression but also decreased p-CREB and c-Fos levels (Fig. 5C, E, F), while the expression of the t-CREB protein (Fig. 5D, F) remained unchanged. We then assessed the expression of KIF1A and CGRP in cells following Kif1a knockdown using IF staining. The results showed that compared with the Control and Control + MS siRNA groups, the Kif1a knockdown group exhibited reduced KIF1A (Fig. 5G) and CGRP immunoreactivity (Fig. 5H). These findings suggested that KIF1A also plays a functional role in this positive feedback loop.

Fig. 5.

Kif1a siRNA downregulates the expression of CGRP, c-Fos, and CREB. A, B Screening of Kif1a-targeting sequences in N2A cells. Representative WB images and qPCR results showing that sequence 2 had the most pronounced inhibitory effect. Statistical significance was analyzed using one-way ANOVA (*p < 0.05, **p < 0.01, and ***p < 0.001 compared with the control; n = 6). C–D Representative WB images showing the validation of the selected Kif1a siRNA in N2A cells, where it effectively suppressed the expression of CGRP, c-Fos, p-CREB (C) and t-CREB (D). E Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to GAPDH. F Ratio of p-CREB to t-CREB. (n = 6 mice per group). Statistical significance was analyzed using one-way ANOVA (*p < 0.05, **p < 0.01, and ***p < 0.001 compared with the control; n = 6). G, H Images of IF staining for KIF1A (G) and CGRP (H) in N2A cells. Nuclei were labeled with DAPI (blue); Scale bars: 50 μm. All the data are presented as the means ± SEMs

p-CREB directly activates Kif1a transcription and regulates CGRP synaptic localization via KIF1A

Double IF staining revealed that KIF1A was predominantly expressed in NeuN-positive neurons within the SP5C, with minimal expression in Iba1-positive microglia and GFAP-positive astrocytes, indicating that migraine-induced KIF1A activation is largely neuron-specific (Fig. 6A). ChIP‒qPCR demonstrated significant enrichment of the Kif1a promoter fragment containing the CREB binding motif in DNA immunoprecipitated with p-CREB antibody, with higher amplification in the NTG group compared to VEH controls (Fig. 6B). Dual-luciferase reporter assays further confirmed that CREB overexpression markedly enhanced Kif1a promoter activity, establishing that CREB directly activates Kif1a transcription (Fig. 6C). Double IF staining also showed colocalization of p-CREB and KIF1A in SP5C neurons (Fig. 6D). To investigate the functional consequences of KIF1A upregulation, co-immunoprecipitation assays demonstrated a physical association between KIF1A and CGRP in SP5C tissue (Fig. 6E), suggesting a role for KIF1A in CGRP-containing vesicle trafficking. Consistently, WB analysis of synaptosome fractions prepared using a commercial extraction kit showed that β-actin levels were comparable across total protein fractions, whereas actin was barely detectable in the synaptosome fractions, confirming the success of the fractionation. Using Synaptophysin and PSD-95 as synaptic loading controls, we observed that CGRP expression was significantly elevated in the NTG + MS-siRNA group compared with the VEH group, while Kif1a-siRNA treatment reduced CGRP expression to near-control levels (Fig. 6F-H). Notably, KIF1A expression exhibited the same pattern as CGRP across experimental groups (Fig. 6F-H).

Fig. 6.

p-CREB directly activates Kif1a transcription and promotes CGRP synaptic localization via KIF1A in SP5C neurons.A Double IF staining showing that KIF1A was predominantly localized in neurons (NeuN-positive), while its expression in microglia (Iba-1-positive) and astrocytes (GFAP-positive) was minimal or nearly absent in the SP5C. B ChIP‒qPCR analysis of p-CREB binding to the Kif1a promoter region in the SP5C (n = 3). C Dual-luciferase reporter assays showing that CREB overexpression markedly increased Kif1a promoter activity. D IF results showing that p-CREB and KIF1A were coexpressed in the mouse SP5C. Scale bars: 100 μm. E Co-immunoprecipitation assay showing the physical association between KIF1A and CGRP in SP5C tissue lysates. Input and IgG controls are included to confirm specificity. F Representative WB images showing the total and synaptosome protein expression of CGRP and KIF1A in SP5C tissues. G Quantification of CGRP and KIF1A protein levels normalized to β-actin (n = 3 mice per group). H Quantification of CGRP and KIF1A protein levels normalized to Synaptophysin (n = 3 mice per group). Statistical analysis: one-way ANOVA for (G, H). *p < 0.05, **p < 0.01, **p < 0.001 vs. VEH group; ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 vs. NTG + MS-RNA group. All data are presented as means ± SEM

Intrathecal injection of Kif1a SiRNA alleviates NTG-Induced nociceptive hypersensitivity

Next, we used the Kif1a siRNA sequence selected through in vitro screening and administered it to mice via intrathecal injection to determine whether it could alleviate migraine-like symptoms. Based on our previous experience with intrathecal administration, Kif1a siRNA was injected 30 min prior to NTG administration. In contrast to the earlier experiments involving i.p. injection of CREB agonists and antagonists, increases in the thermal and mechanical nociceptive thresholds of the hind paws were observed only after the third intrathecal injection of the Kif1a siRNA (on day 5), with significant increases evident by day 9 (Fig. 7A, B). Moreover, periocular allodynia began to improve by day 7, with this effect persisting through day 9 (Fig. 7C).

Fig. 7.

Kif1a siRNA attenuated NTG-induced hyperalgesia and reduced the upregulation of CGRP, c-Fos, and p-CREB expression. A–C NTG induced mechanical allodynia (decreased HML and HMT) and thermal hyperalgesia (decreased PMT), whereas compared with NTG + MS-RNA, Kif1a siRNA attenuated these nociceptive behaviors (n = 8 mice). D–E Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (D), and t-CREB (E) in SP5C tissues. F Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to those of GAPDH. (G) Ratio of p-CREB to t-CREB. (n = 6 mice per group). H, J IF staining for CGRP (H) and c-Fos (J) in the SP5C. Nuclei were labeled with DAPI (blue); Scale bars: 200 μm (H) and 100 μm (J). I, K Quantification of the fluorescence intensity of CGRP (H) and c-Fos (J) in each group (n = 4 per group). Statistical analysis: Two-way ANOVA for (A–C); one-way ANOVA for (F, G, I, K). *p < 0.05, **p < 0.01, **p < 0.001 vs. VEH + MS-RNA group; ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 vs. NTG + MS-RNA group. All the data are presented as the means ± SEMs

WB analyses demonstrated that Kif1a siRNA significantly reduced CGRP expression (Fig. 7D, F). Furthermore, Kif1a inhibition decreased p-CREB levels, while t-CREB protein expression remained unchanged (Fig. 7E, G). Intrathecal injection of Kif1a siRNA into male mice also significantly suppressed the NTG-induced upregulation of c-Fos (Fig. 7D, F). IF staining of the SP5C region corroborated these findings: compared to the MS siRNA-treated group, Kif1a siRNA markedly reduced CGRP (Fig. 7H, I) and c-Fos (Fig. 7J, K) expression. In contrast, the MS siRNA group showed no significant changes in the expression of these proteins (Fig. 7J, K).

Overexpression of Kif1a upregulates CGRP and Kif1a expression and is inhibited by a CGRP inhibitor

We overexpressed Kif1a in N2A cells to investigate the function of Kif1a and its potential interaction with CGRP signaling. Kif1a overexpression led to significant increases in both KIF1A and CGRP protein levels compared to Control cells (Fig. 8A–D). This coordinated increase indicates a potential positive feedback loop between Kif1a and CGRP signaling. The CGRP receptor antagonist Olcegepant exhibited a dose-dependent and specific effect, as demonstrated by appropriate control treatments (Fig. 8E–H). Importantly, the upregulation of CGRP and KIF1A induced by Kif1a overexpression was significantly attenuated by cotreatment with the Olcegepant (Fig. 8I–L). Cellular IF showed increased fluorescence expression of KIF1A (Fig. 10A) and CGRP (Fig. 10B) following Kif1a overexpression compared to the Control group, which was inhibited by Olegepant (Fig. 10A, B). These findings demonstrated that Kif1a overexpression increases the expression of both CGRP and itself, and this effect is functionally dependent on CGRP receptor signaling.

Fig. 8.

Kif1a overexpression increases the expression of CGRP and CREB, which is blocked by a CGRP receptor antagonist. A-B Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (A), and t-CREB (B) in N2A cells following transfection with the Kif1a overexpression plasmid. C Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to those of GAPDH. D Ratio of p-CREB to t-CREB. (n = 6). E-F Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (E) and t-CREB (F) in N2A cells following treatment with different concentrations of the CGRP receptor antagonist Olcegepant. G Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to GAPDH. H Ratio of p-CREB to t-CREB. (n = 6). (I-J) Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (I), and t-CREB (J) in N2A cells. K Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to those of GAPDH. L Ratio of p-CREB to t-CREB. (n = 6). Statistical significance: All the data were analyzed using one-way ANOVA. **p < 0.01 and ***p < 0.001 compared with the control group; ^##p < 0.01 and ^###p < 0.001 compared with the pcDNA.3.1-Kif1a group. All the data are presented as the means ± SEMs

Fig. 10.

FSK stimulation and Kif1a overexpression upregulate the expression of CGRP and KIF1A. A, B Representative IF images showing KIF1A (A) and CGRP (B) expression in N2A cells following FSK stimulation. Nuclei were counterstained with DAPI (blue). Scale bars: 50 μm. C, D Representative IF images showing KIF1A (C) and CGRP (D) expression in N2A cells following Kif1a overexpression. Nuclei were counterstained with DAPI (blue). Scale bars: 50 μm

CGRP receptor Blockade disrupts the FSK-Induced p-CREB/KIF1A axis

We delineated the role of CGRP signaling in FSK-mediated transcriptional activation by first characterizing the time-dependent effects of FSK (10 µM) on the levels of key signaling effectors. FSK stimulation significantly upregulated p-CREB, c-Fos, CGRP, and KIF1A protein expression in a time-dependent manner, with the peak increase observed at 12 h for p-CREB, c-Fos, and KIF1A and sustained elevation of CGRP expression being observed (Fig. 9A-D). Critically, pharmacological blockade of the CGRP receptor with Olcegepant (50 nM) abolished the FSK-induced activation of this signaling cascade. Cotreatment with Olcegepant markedly attenuated the FSK-triggered phosphorylation of CREB and suppressed the upregulation of c-Fos, CGRP, and KIF1A expression (Fig. 9E-H). T-CREB levels remained unchanged across all groups, confirming the specificity of phospho-CREB inhibition (Fig. 9E, H). FSK treatment increased KIF1A (Fig. 10C) and CGRP (Fig. 10D) fluorescence expression detected by cellular IF compared to the Control group, an effect inhibited by Olegepant (Fig. 10C, D). These results indicate that CGRP receptor signaling is essential for the FSK-induced activation of the p-CREB/c-Fos pathway and subsequent KIF1A expression.

Fig. 9.

An agonist of CREB increases the expression of CGRP and CREB, an effect that is blocked by a CGRP receptor antagonist. A-B Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (A) and t-CREB (B) in N2A cells following treatment with the CREB agonist FSK for different durations. C Quantification of CGRP, c-Fos, and KIF1A protein levels normalized to those of GAPDH. D Ratio of p-CREB to t-CREB. (n = 6). E–F Representative WB images showing the protein expression of KIF1A, c-Fos, p-CREB, CGRP (E), and t-CREB (F) in N2A cells. G–H Quantification of protein levels normalized to those of GAPDH or total CREB (n = 6). Statistical significance was analyzed using one-way ANOVA (**p < 0.01 and ***p < 0.001 compared with the control group; ^##p < 0.01 and ^###p < 0.001 compared with the FSK group; n = 6). Statistical significance: All the data were analyzed using one-way ANOVA. **p < 0.01 and ***p < 0.001 compared with the control group; ^###p < 0.001 compared with the FSK group. All the data are presented as the means ± SEMs

Discussion

This study identifies a self-amplifying CREB–KIF1A–CGRP signaling circuit that functions as a pathological engine driving central sensitization in CM for the first time. Three findings support this loop: (i) p-CREB directly binds the Kif1a promoter to drive its expression; (ii) KIF1A regulates CGRP vesicle transport noncanonically; and (iii) CGRP activates neuronal receptors, triggering CREB phosphorylation via cAMP/Protein Kinase A (PKA).

As part of the trigeminovascular system, the SP5C is a key central hub in the initiation and progression of migraine [23, 24]. Elevated CGRP levels and c-Fos expression are hallmarks of the mouse migraine model used here. Thus, the NTG model is a valuable tool for evaluating the efficacy of novel therapeutics targeting CGRP in migraine sufferers [6, 25]. As a typical downstream transcription factor of the cAMP/PKA pathway, CREB is widely expressed in the brain and spinal dorsal horn and is upregulated in headache-related regions [26–28]. Previous studies have shown that CGRP receptor antagonists reduce CREB expression [29], while cAMP/CREB signaling can in turn promote CGRP expression [30, 31], suggesting the existence of a positive feedback loop: injury- or inflammation-induced CGRP release activates its receptor, elevates intracellular cAMP levels, and phosphorylates CREB, which then drives the transcription of the CGRP gene and other pronociceptive targets, thereby amplifying and sustaining nociceptive signaling. Our previous work demonstrated that the axonal motor protein KIF1A participates in pain processing, as Kif1a-deficient mice exhibit hypoalgesia [14, 32]. Given that KIF1A transports dense-core vesicles containing CGRP precursors along microtubules [33, 34], we hypothesize that CREB activation not only promotes CGRP synthesis but may also upregulate KIF1A expression, thereby enhancing the biogenesis, storage, and axonal transport of CGRP-loaded vesicles. This mechanism could further reinforce the positive feedback loop, maintaining high levels of CGRP signaling across neuronal networks even in the absence of direct autocrine receptor activation [35, 36].

Using a well-established NTG-induced CM model, we observed that repeated NTG administration led to progressive thermal and mechanical hypersensitivity in both hind paw and periorbital regions, accompanied by increased levels of p-CREB, KIF1A, CGRP, and the immediate early gene c-Fos in the SP5C. These findings suggest the central amplification of nociceptive signaling and implicate the CREB–KIF1A–CGRP axis as a key regulatory pathway in migraine pathophysiology.

FSK is a diterpene produced from the roots of the Indian plant Coleus forskohlii [37]. FSK directly activates adenylate cyclase, an enzyme that generates cAMP from ATP, thereby increasing intracellular cAMP levels and subsequently increasing the expression of PKA/CREB1 [38, 39]. To date, no studies have explored the relationship between FSK and migraine in animal models. In our study, we first administered FSK via i.p. injection and established a dose gradient. Based on the behavioral outcomes, we selected the medium dose for subsequent experiments. The results showed that FSK could induce migraine-like symptoms by activating the CREB signaling pathway and simultaneously upregulate KIF1A expression (Fig. 3). In this study, we did not assess PKA protein levels after FSK administration, as FSK is a well-established activator of the cAMP–PKA pathway. Given the widely accepted upstream mechanism, we focused on downstream effectors, particularly CREB and KIF1A, and explored their roles in migraine-related behaviors.

The potent and selective CREB inhibitor 666 − 15 decreases CREB activation [40]. Following the activation of CREB to assess migraine-like behaviors and the CREB–KIF1A–CGRP loop in mice, we next investigated whether inhibiting CREB could alleviate these effects. We administered the CREB inhibitor 666 − 15 via i.p. injection to determine its impact on migraine-like responses and the expression levels of KIF1A and CGRP. As predicted, treatment with 666 − 15 significantly attenuated migraine-like symptoms in mice. WB analyses showed that KIF1A and CGRP levels were markedly reduced in the NTG + 666 − 15 group compared with the NTG + VEH group (Fig. 4).

Mechanistically, we showed that p-CREB directly binds to the promoter region of the Kif1a gene in vivo, as demonstrated by ChIP–qPCR, and colocalizes with KIF1A-positive neurons in the SP5C, allowing direct transcriptional regulation (Fig. 6A–B). Consistently, a dual-luciferase reporter assay confirmed that CREB significantly enhances Kif1a promoter activity, further supporting a direct functional interaction between CREB and Kif1a transcription (Fig. 6C). These results are particularly significant as they provide molecular evidence that CREB acts upstream of KIF1A in migraine-associated neuronal activation. Furthermore, we identified neurons, but not microglia or astrocytes, as the primary cellular source of KIF1A in this context, indicating that the CREB–KIF1A axis is neuron-specific (Fig. 6D). Finally, coimmunoprecipitation experiments revealed that KIF1A physically associates with CGRP protein (Fig. 6E), suggesting a direct link between KIF1A-mediated transport machinery and CGRP-containing vesicles in migraine-relevant neurons.

We examined whether KIF1A is necessary for maintaining migraine-like hypersensitivity by performing both in vitro and in vivo knockdown experiments. Silencing of Kif1a in N2A cells resulted in the downregulation of CGRP, p-CREB, and c-Fos, but not t-CREB, highlighting the regulatory role of KIF1A in CREB phosphorylation (Fig. 5). Similarly, the intrathecal administration of Kif1a siRNA to mice significantly alleviated NTG-induced thermal and mechanical hypersensitivity and reduced the expression of CGRP and c-Fos in the SP5C (Fig. 7). To further investigate the synaptic distribution of CGRP, we isolated synaptosomal fractions from the SP5C using a commercial extraction kit, with successful enrichment confirmed by the absence of β-actin (non-synaptic contaminant) and presence of synaptic markers synaptophysin and PSD-95—ensuring precise assessment of nerve terminal localization. Our results demonstrated that KIF1A critically directs CGRP to synaptic sites, as Kif1a knockdown induced a parallel reduction of both KIF1A and CGRP specifically in synaptosomal compartments, establishing that KIF1A facilitates anterograde axonal transport and presynaptic accumulation of CGRP-containing vesicles rather than merely modulating total CGRP expression. Conversely, synaptosomal CGRP levels were markedly elevated in the NTG + MS-siRNA group compared to VEH controls, reflecting enhanced trafficking or release during migraine sensitization—consistent with CGRP’s established pathophysiological role (Fig. 6F-H). Collectively, these findings mechanistically link KIF1A-dependent vesicular transport to synaptic CGRP targeting and nominate KIF1A as a therapeutic target for CGRP-driven pain disorders.

Overexpression of Kif1a in vitro resulted in the marked upregulation of both CGRP and KIF1A itself, an effect that was attenuated by cotreatment with the CGRP receptor antagonist Olcegepant (Fig. 8). This finding suggests that CGRP signaling contributes to Kif1a transcription or protein stability, reinforcing the existence of a self-sustaining CREB–KIF1A–CGRP circuit. Supporting this finding, pharmacological blockade of CGRP receptors abolished the FSK-induced activation of CREB, c-Fos, and KIF1A, further indicating that CGRP signaling is essential for sustaining the CREB-dependent transcriptional cascade. This cascade may underlie the transition from episodic to chronic migraine by establishing a molecular memory of pain through persistent CREB activation and CGRP upregulation.

The mechanistic primacy of this axis is underscored by its vulnerability to disruption at multiple points: CGRP receptor blockade abolished the effects of KIF1A overexpression (Fig. 8), and pharmacological inhibition with Olcegepant not only suppressed migraine-like behaviors but also eliminated both KIF1A-overexpression- and FSK-induced upregulation of p-CREB, KIF1A, and CGRP (Fig. 9). These results indicate that CGRP receptor activity serves as a critical amplifier and permissive regulator within the CREB–KIF1A signaling axis. Beyond clarifying migraine pathophysiology, this insight reframes therapeutic strategy. While current CGRP-targeted biologics are often regarded as peripheral neuropeptide scavengers, our data may reveal a central mechanism: by interrupting CGRP–CGRPR signaling, they deprive the CREB–KIF1A axis of its essential feedback stimulus, dismantling the sensitization machinery at its amplification point (Figs. 8 and 9). Moreover, our findings nominate KIF1A as a high-value druggable node—its neuron-restricted expression (Fig. 6), positioning downstream of CREB but upstream of CGRP, and functional indispensability (Figs. 6, 7 and 8) make it an attractive target for breaking migraine chronification without the cognitive risks associated with global CREB inhibition. The efficacy of intrathecal KIF1A silencing (Fig. 7) provides preclinical proof of concept for KIF1A inhibitors, representing a therapeutic avenue with potential superiority through proximal intervention within the feedback loop.

This study has several limitations. First, although the NTG-induced migraine model reflects key features of migraine, it does not fully replicate the complexity of human pathophysiology, limiting translational generalizability. Second, although our biochemical data indicate a role for KIF1A in CGRP transport, we were unable to conduct real-time fluorescence imaging of KIF1A-mediated vesicle trafficking. Third, given KIF1A’s broad role in axonal transport, its long-term inhibition may have unintended effects, which warrant further investigation. Last, all data were derived from murine models, and validation in human tissue will be essential to assess clinical relevance.

Conclusions

In conclusion, we delineate a CREB→KIF1A→CGRP→CREB positive feedback circuit that operates as a nonlinear pathological oscillator in chronic migraine. KIF1A emerges not as a passive cargo transporter but as an active signaling hub, converting transcriptional activation into neuropeptide dysregulation while receiving reciprocal reinforcement from CGRP. This self-perpetuating architecture explains why migraine can evolve into an autonomous entity largely independent of initial triggers, and why disrupting key components (e.g., CREB, KIF1A, or CGRP) silences the circuit. Targeting this loop represents a promising strategy to halt migraine chronification.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ANOVA

Analysis of variance

ChIP

Chromatin immunoprecipitation

CGRP

Calcitonin gene-related peptide

CREB

cAMP response element-binding protein

CM

Chronic migraine

FSK

Forskolin

HML

Hindpaw thermal latency

HMT

Hindpaw mechanical threshold

IF

Immunofluorescence

KIF1A

Kinesin family member 1A

N2A

Neuro-2a

NTG

Nitroglycerin

PBS

Phosphate-buffered saline

PKA

Protein Kinase A

PMT

Periorbital mechanical threshold

qRT‒PCR

Quantitative real-time PCR

VEH

Vehicle

WB

Western blot

WT

Wild-type

SP5C

Spinal trigeminal nucleus caudalis

Author contributions

JW, YP and SYM conceived of and designed the study and drafted a significant portion of the manuscript or figures. ZLX and CMT acquired and analyzed the data and helped with the writing and revision of the manuscript. YY and DM contributed significantly to the revision of the final manuscript. All the authors have read and approved the final manuscript.

Funding

This work was supported by grants from the General Program of the Jilin Provincial Natural Science Foundation (Medical Science Field) (No. YDZJ202501ZYTS084); The Jilin Provincial Health Science and Technology Capacity Enhancement Program Project (No. 2024A015); National Key Research and Development Program of China (No. 2023YFC2508701); Jilin Province Natural Science Foundation of China (No. JLSRCZX2025045); Jilin Province Natural Science Foundation of China (No. YDZJ202201ZYTS100); and Technology Innovation 2030 Major Projects of China (No. 2022ZD0211605).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The study protocol received prior approval from the Institutional Animal Care and Use Committee of Jilin University.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.