Changes in the gene expression profile during spontaneous migraine attacks

DOI: https://doi.org/10.1038/s41598- 021- 87503- 5

Abstract-Summary Gene expression fluctuates according to environmental and endogenous events and therefore, we hypothesized that changes in RNA expression during and outside a spontaneous migraine attack exist which are specific to migraine.

Twenty-seven migraine patients were assessed during a spontaneous migraine

attack, including headache characteristics and treatment effect.

RNA-Sequences were analyzed at gene level (differential expression analysis)

and at network level, and genomic and transcriptomic data were integrated.

We found 29 differentially expressed genes between ‘attack’ and ‘after treat- ment’, after subtracting non-migraine specific genes, that were functioning in fatty acid oxidation, signaling pathways and immune-related pathways.

We uniquely investigated intra-individual changes in gene expression during a

migraine attack.

We revealed both genes and pathways potentially involved in the pathophysiol-

ogy of migraine and/or migraine treatment.

Extended: Our study suggests several molecular mechanisms involved in migraine pathophysiology and/or its treatment with sumatriptan, with an important role of mitochondria, Notch signaling, ion channels, the immune system and previ- ous findings suggesting platelet abnormalities.

Introduction A few gene expression studies indicated a difference between migraine patients outside of migraine attack and healthy controls [76–78], but these studies were explorative and after correction for multiple testing, we have recently showed that there is no distinct difference outside of attack [79].

Only one study has investigated gene expression during migraine attack [80]. This study compared the migraine patients during attack to healthy controls which is not an optimal design because of huge interindividual gene expression variability.

Given this study design it was not possible to establish whether the gene expres- sion was altered due to attack or to other clinical characteristics with a reasonable precision although some interesting findings, such as the altered expression of platelet- related genes, supported further study of RNA expression in migraine.

We here investigated the intra-individual changes in gene expression during a migraine attack which lasts between four and 72 h, enabling a paired-sample design. The specific aim of the present study was to investigate gene expression profiles using RNA-sequencing of migraine patients during a migraine attack, two hours after receiving acute medication, and outside migraine attack.

234

2 Mechanisms

Methods Differential gene expression analysis was performed as a paired sample design, extracting samples of two time points at a time.

To exploit the temporal design, we performed a likelihood ratio test with the full model including ‘Time’ as covariate using the samples taken during attack (A), after treatment (B), and headache-free (C), and the reduced model omitted ‘Time’ as covariate.

Plotting the overlap of genes in clusters in the two subnetworks (during attack vs after treatment and headache-free vs after cold-pressor test) resulted in a sigmoid curve where we then picked the switching point to ensure a balance in optimization of the clustering score.

In the preliminary analysis, the absence of significant DE genes in ‘during attack’ vs ‘headache-free’ indicated that we did not have enough power to detect changes with this sample size over a longer time frame.

Results Although the cortisol levels did not significantly decrease after treatment, we did see one of the DE genes to be significantly correlated with cortisol levels: DDIT4 (r = 0.78, Padj = 9.11 × 10–10).

We did not see an overall differential expression within the pathway, (p = 0.62),

but one gene was significantly DE: BRAF (p = 3.79 × 10–4).

If a change in gene interaction during migraine attack (from network ‘during attack’ [A] to network ‘after treatment’ [B]) led to a significant change in pathway composition, based on gene ontology (GO) enrichment, we considered the path- ways affected by, or interacting with, the migraine attack or its treatment.

Of the 41 genes targeted by risk variants from the migraine GWAS, 24 were pres- ent in the differential network, for example ECM1 and MPPED2 in cluster 9, with GO-enrichment of ‘detection of stimulus involved in sensory perception’ during attack (A).

Discussion Using RNA-Sequencing we investigated changes in gene expression during a migraine attack and revealed genes and pathways that are potentially involved in the underlying mechanisms of migraine.

Among the up- and down-regulated genes during the migraine attack, treated with sumatriptan, were genes functioning in fatty acid oxidation (CPT1A, SLC25A20 and ETFDH), signaling pathways such as Notch signaling (MAML2, ADAM15 and ADAM17) and immune-related pathways (CARD9, SH2D2A, CD300C).

We showed that the paired-sample design gives the power to reveal molecular mechanisms involved in a migraine attack and its treatment and that the cold pressor test, importantly, excluded changes due to a general stress/pain response.

Our study suggests several molecular mechanisms involved in migraine patho- physiology and/or its treatment with sumatriptan, with an important role of mito- chondria, Notch signaling, ion channels, the immune system and previous findings suggesting platelet abnormalities.

2.1 Genetics

235

Acknowledgement A machine generated summary based on the work of Kogelman, Lisette J.  A.; Falkenberg, Katrine; Buil, Alfonso; Erola, Pau; Courraud, Julie; Laursen, Susan Svane; Michoel, Tom; Olesen, Jes; Hansen, Thomas F. 2021 in Scientific Reports.

The NRP1 migraine risk variant shows evidence of association with menstrual migraine