Maladaptive Activation of Nav1.9 Channels by Nitric Oxide Causes Triptan-Induced Medication Overuse Headache

ARTICLE

https://doi.org/10.1038/s41467-019-12197-3 OPEN Maladaptive activation of Nav1.9 channels by nitric oxide causes triptan-induced medication overuse headache

Caroline Bonnet1, Jizhe Hao1, Nancy Osorio1, Anne Donnet2, Virginie Penalba1, Jérôme Ruel1 & Patrick Delmas1

Medication-overuse headaches (MOH) occur with both over-the-counter and pain-relief medicines, including paracetamol, opioids and combination analgesics. The mechanisms that 1234567890():,; lead to MOH are still uncertain. Here, we show that abnormal activation of Nav1.9 channels by Nitric Oxide (NO) is responsible for MOH induced by triptan migraine medicine. Deletion of the Scn11a gene in MOH mice abrogates NO-mediated symptoms, including cephalic and extracephalic allodynia, photophobia and phonophobia. NO strongly activates Nav1.9 in dural afferent neurons from MOH but not normal mice. Abnormal activation of Nav1.9 triggers CGRP secretion, causing artery dilatation and degranulation of mast cells. In turn, released mast cell mediators potentiates Nav1.9 in meningeal nociceptors, exacerbating inﬂammation and pain signal. Analysis of signaling networks indicates that PKA is downregulated in trigeminal neurons from MOH mice, relieving its inhibitory action on NO-Nav1.9 coupling. Thus, anomalous activation of Nav1.9 channels by NO, as a result of chronic medication, promotes MOH.

1 Aix-Marseille-Université, CNRS, Laboratoire de Neurosciences Cognitives, UMR 7291, CS8011, Bd Pierre Dramard, 13344 Marseille, France. 2 Centre d’évaluation et de traitement de la douleur, Hôpital de la Timone, Marseille, France. Correspondence and requests for materials should be addressed to P.D. (email: [email protected])

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3

hronic headache encompasses many different headache by NO as a central determinant of triptan-induced MOH and Cdiagnoses and include chronic migraines, chronic tension- pave the way for the development of mechanism-based treatment type headaches, medication-overuse headaches (MOH), strategies that can improve the management of primary and other types of daily persistent headaches. Migraine, a fre- headaches. quently incapacitating neurovascular disorder, affects hundreds of 1 millions of individuals worldwide . It is characterized by a severe, Results debilitating and throbbing unilateral headache accompanied by a Nav1.9 is expressed in meningeal nociceptors. Immunostaining host of neurological symptoms including hypersensitivity to of mouse TG cryosections indicated that Nav1.9 is expressed in visual and auditory stimulation, nausea and vomiting, and a 32% of TG neurons (Fig. 1a). Nearly all immunopositive neurons variety of autonomic, cognitive and motor disturbances2,3. (96%, n = 174) exhibited small-diameter or medium-diameter Current antimigraine drugs target trigeminovascular 5-HT 1B/ cell bodies (<27 µm, average largest diameter ⌀ = 17.4 ± 0.3 µm) , 5-HT , and CGRP receptors. These different antimigraine 1D 1F (Fig. 1b), while a minority of them (3.8%, n = 7 out of 181) medications induce adverse side effects, including MOH, which is displayed large cell bodies (⌀ ≥ 27 µm). Small peripherin-positive a worldwide health problem with a prevalence range of 1–2% fibers showed features of apposition with meningeal arteries in with a 3:1 female to male ratio. MOH is a severe form of headache whole mount of mouse dura mater (Supplementary Fig. 1A). where the patients are prone to develop primary headaches with Dual-labeling showed that 89% (n = 124/139) and 91% (n = 101) unsuccessful treatments. Patients with migraine or tension-type of peripherin-positive meningeal fibers were immunoreactive for headaches have a higher potential for developing MOH than Nav1.9 and CGRP, respectively (Fig. 1c, d), suggesting that other primary headaches. MOH does not develop in persons Nav1.9 and CGRP co-distribute in a large proportion of menin- without a history of headache when medication is being used for geal fibers. Double-labeling for Nav1.9 and CGRP could not be other conditions, such as inflammatory diseases. In addition, made due to the different fixation conditions of primary anti- virtually all medication for headaches may lead to MOH, bodies. However, the quasi-totality of β-gal -positive TG neurons including opioids, ergotamine, butalbital-containing medicine, exhibited CGRP staining in cryosections from Scn11a-GAL triptans or a combination thereof. Thus, MOH is generated in reporter transgenic mice (n = 63/64) (Supplementary Fig. 1B). headache-prone persons by the interaction between headache Few Nav1.9-positive meningeal fibers were also found to be medication and pre-existing headache disorder pathways. immunoreactive for NF200, possibly corresponding to some The precise mechanisms that lead to MOH development are lightly myelinated sensory fibers (Supplementary Fig. 1C, D). A largely unknown. However, multiple factors may be implicated, majority (57%) of retrogradely labeled (DiI) dural afferent neu- including genetic predisposition, sensitization within the rons was found to express Nav1.9 current using patch clamp trigeminal (TG) system, abnormal cortical pain processing + recording (Fig. 1e, g). Consistently, 54% of DiI -dural afferent and decreased antinociceptive activity of the supraspinal struc- – neurons from Scn11a-GAL reporter transgenic mice exhibited β- tures4 7. gal enzymatic activity (Fig. 1f, g). Together, these data provide Multiple studies indicate that migraine medication induces evidence that about half of dural afferent neurons expresses sensitization of peripheral and central pain pathways. For functional Nav1.9 channels. instance, chronic use of opioids and triptans in animals has been shown to increase the level of calcitonin gene-related peptide (CGRP), which is involved in neurogenic inflammation and Scn11a gene inactivation abrogates NO-induced allodynia in headache pain8,9. These animals develop a persistent hypersen- MOH mice. To probe the role of Nav1.9 channels in MOH, we sitivity or latent sensitization to provocative triggers, such as developed a model of MOH in Nav1.9−/− mice and their wild- environmental stress stimulus and the well-known human type (WT) littermates using triptan medicines and assessed migraine trigger nitric oxide (NO). This latent sensitization per- quantitatively behavioral correlates of headache and migrainous sists long after discontinuation of drug administration and pro- symptoms, including generalized allodynia, photophobia, and duce a state of generalized cutaneous allodynia that was detected phonophobia. Osmotic minipump infusion of sumatriptan in in periorbital regions and hind paw. WT mice for 6 days (0.6 mg/kg/day) produced a time- To probe molecular mechanisms that lead to MOH, we dependent reduction in mechanical withdrawal thresholds of developed a MOH mouse model based on the sustained admin- the hind paw relative to saline-treated WT mice (Fig. 2a). istration of sumatriptan, a 5-HT receptor agonist selective for 5- Recovery of paw sensory thresholds to sumatriptan pre- HT1D and 5-HT1B subtypes. We applied the MOH model to infusion values occurred within 18–20 days after minipump transgenic mice for studying the role of the nociceptor-specific implantation (Fig. 2a). Latent sensitization to the migraine voltage-gated Nav1.9 channel in headache. The function of this trigger NO was tested at day 21, once sensory thresholds were channel in the TG pain pathway is still poorly understood10–12. returned to pre-sumatriptan baseline levels (Fig. 2b). Injection Nav1.9 channel is known to generate a persistent, tetrodotoxin of the NO donor sodium nitroprusside (SNP, 0.03 mg/kg) into (TTX)-resistant Na+ current that promotes sustained neuronal the loose skin over the neck evoked strong extracephalic tactile activity in dorsal root ganglion (DRG) neurons13–17. Nav1.9 is allodynia in sumatriptan-treated WT mice compared with known to contribute to both inflammatory18–22 and saline-treated WT mice (Fig. 2b). Heightened SNP-induced allo- neuropathic22,23 somatic pain in animal models, and variants of dynia in sumatriptan-treated WT mice was still observable up to the Scn11a gene encoding Nav1.9 in humans lead to congenital 46 days after minipump implantation (Supplementary Fig. 2A). insensitivity to pain and to painful syndromes24–27. Heightened mechanical allodynia induced by SNP injection at day Using molecular, electrophysiological, and behavioral approa- 21 was absent in sumatriptan-treated Nav1.9−/− mice (Fig. 2d). ches, we show that mice chronically treated with sumatriptan Moreover, saline-treated Nav1.9−/− and WT mice displayed display increased responsiveness of Nav1.9 to NO, leading to similar SNP-induced (basal) allodynia (Fig. 2b, d), indicating that headache/migraine-like symptoms including generalized allody- Nav1.9 contributes to the SNP-induced heightened allodynia in nia, photophobia, and phonophobia. MOH mice show deficit in sumatriptan-treated WT mice, but plays no apparent role in SNP- PKA-mediated inhibition of NO–Nav1.9 coupling, causing induced basal allodynia in saline-treated animals. hyperactivity of meningeal nociceptors and inflammation in the Because females have increased risk of developing migraine meninges. Thus, our data identify abnormal activation of Nav1.9 and MOH, we tested whether infusion of sumatriptan for 6 days

2 NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3 ARTICLE

a Peripherin Nav1.9 Merge b Peripherin 80 Peripherin + Nav1.9 60 Nav1.9 40

20 Number of cells ∅ (μm) 100 μm

7.5 c Peripherin Nav1.9 Merge 10.513.516.5 19.522.5 25.528.531.534.5 37.5 40.5 e Nav1.9

250 pA 50 μm 40 μm 50 ms d CGRP Peripherin Merge f g Nav1.9 current Scn11a-driven β-gal Scn11a-βgal 80

60 30 285

20 Dil-neurons (%) 100 μm 25 μm 0

Fig. 1 Nav1.9 is expressed in meningeal nerve ﬁbers and dural afferent neurons. a Cryosections of a mouse TG were double-labeled for peripherin and Nav1.9. Images are projections of seven consecutive optical sections spanning 9 µm. Rightmost panel: merged image. b Histogram showing the distribution of TG neurons immunopositive for peripherin (green), Nav1.9 (red), or both (yellow) as a function of cell body diameter (⌀). Bin size = 3 µm. c Mouse whole-mount of dura mater was double-labeled for peripherin and Nav1.9. Images are projections of 16 consecutive optical sections spanning 22 µm. Rightmost panel: merged image. d Immunostaining for CGRP and peripherin in mouse dura mater. Images are projections of 21 consecutive optical sections spanning 23 µm. Rightmost panel: merged image. e Left panel: TG neurons cultured 2 days after DiI application through a cranial window in the parietal bone. Right panel: recording of Nav1.9 current in the retrogradely labeled TG neuron indicated by an arrow. The current was evoked by 100 ms-voltage steps from −80 to −5 mV from a Vh of −100 mV. CsF-containing patch pipette solution. f Image from a Scn11alacZ mouse showing β-galactosidase expression (dark dots) in DiI+ afferent neurons (red) from the TG. g Histogram showing the percentage of DiI+ afferent neurons exhibiting Nav1.9 current (patch clamp) or β-galactosidase (staining)

(0.6 mg/kg/day) produced mechanical hypersensitivity and latent Lack of NO-induced visual and auditory symptoms in Nav1.9 sensitization to NO in WT female mice as observed for the KO mice. Besides pain, disabling symptoms of MOH often opposite gender. Chronic sumatriptan produced a strong include photophobia and/or phonophobia28. To evaluate light- reduction in mechanical withdrawal thresholds of the hind paw aversive behavior (photophobia) of mice after SNP injection, we relative to saline-treated WT female mice and to sumatriptan- used the light–dark transition test29. Two hours after SNP treated Nav1.9−/− female mice (Supplementary Fig. 2B). Injec- injection, sumatriptan-treated WT mice spent significantly more tion of SNP (0.03 mg/kg) at day 21 once sensory thresholds had time in the dark chamber relative to saline-treated WT mice returned to pre-sumatriptan baseline values caused significantly (Fig. 3a). Sumatriptan-treated Nav1.9−/− mice challenged with stronger mechanical allodynia in sumatriptan-treated WT female SNP showed no signs of photophobia as the animals spent no mice compared to saline-treated WT female mice (Supplementary more time in the dark box than saline-treated WT or Nav1.9−/− Fig. 2C). SNP-induced heightened mechanical allodynia was mice treated likewise (Fig. 3a). absent in sumatriptan-treated Nav1.9−/− female mice (Supple- We further quantified the sound sensitivity (phonophobia) of mentary Fig. 2D), reaching similar amplitude to that caused by mice following SNP injection by measuring the intensity of sound SNP in saline-treated Nav1.9−/− female mice (Supplementary required to induce the acoustic startle reflex (ASR). We exposed Fig. 2D). This series of experiments shows that Nav1.9, as mice to pseudo-random order of 100 ms-long white noise bursts observed in male mice, had no role in SNP-induced allodynia in ranging from 60 to 120 dB SPL at 10 kHz. The ASR threshold saline-treated animals, but contributes to the heightened SNP of saline-treated WT mice injected with SNP was 97 ± 2 dB SPL allodynia in sumatriptan-treated female mice. (n = 7), which was significantly different than that of sumatriptan- SNP injection at day 21 was also found to reduce mean treated WT mice, which showed a lower ASR threshold of 90 ± 1.5 periorbital von Frey thresholds (periorbital allodynia) in dB SPL (n = 6) (Fig. 3b). Because one decibel equals 10 times the sumatriptan-treated WT male mice compared with saline- common logarithm of the power ratio, a decrease in ~10 dB SPL in treated WT male mice (Fig. 2e). Deletion of Nav1.9 significantly ASR corresponds to a 10-fold increase in sound sensitivity. SNP attenuated the SNP-induced periorbital allodynia in sumatriptan- induced no changes in ASR threshold in sumatriptan-treated treated animals (Fig. 2e). Altogether, these data indicate that Nav1.9−/− mice compared to saline-treated Nav1.9−/− mice Nav1.9 contributes to NO-induced cephalic and extracephalic (Fig. 3b). We then examined the prepulse inhibition (PPI) of the tactile allodynia in MOH mice. ASR. This response provides an operational measure of

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3

Osmotic minipump Sumatriptan or saline solution SNP injection

d–2 d0 d6 d21 Time (days)

a b SNP or vehicle inj. 1.2 Infusion of sumatriptan Day 21 or saline solution 1.2

1.0 1.0 e

0.8 0.8 WT saline (n = 8) 0.6 0.6 WT suma. (n = 6) Nav1.9–/– suma. (n = 6) 0.4 0.4 Day 21 WT saline, vehicle (n = 4) 1.4 WT saline, SNP (n = 12) 0.2 WT saline (n = 12) 0.2 WT suma., SNP (n = 15) 1.2 Normalized tactile withdrawal threshold Normalized tactile withdrawal WT sumatriptan (n = 15) threshold Normalized tactile withdrawal 0.0 0.0 –2 024681012 14 16 18 20 010 2 4 6 8 1.0 Time (day) Time (h) 0.8

cdSNP inj. 0.6 1.2 Infusion of sumatriptan 1.2 Day 21 0.4 1.0 1.0 Normalized periorbital threshold 0.2 0.8 0.8 0.0 0 24 6 0.6 0.6 Post-SNP injection time (h)

0.4 0.4 WT suma. (n = 15) Nav1.9–/– saline (n = 8) 0.2 WT (n = 16) 0.2 –/– –/– Nav1.9 suma. (n = 8) Normalized tactile withdrawal threshold Normalized tactile withdrawal Normalized tactile withdrawal threshold Normalized tactile withdrawal Nav1.9 (n = 18) 0.0 0.0 –2024681012 14 16 18 20 0 2 4 6 8 10 Time (day) Time (h)

Fig. 2 Deletion of Nav1.9 prevents NO-induced generalized allodynia in MOH mice. a Infusion of sumatriptan (0.6 mg/kg/day), but not saline solution (0.9%), decreases withdrawal thresholds to tactile stimuli applied to the hind paws of WT mice. Hind paw withdrawal threshold was tested using von Frey filaments (inset). **p < 0.01, ***p < 0.001 compared to saline with Mann–Whitney non-parametric test. Top inset: schematic of mouse treatment over time. b Changes in mechanical withdrawal thresholds of the hind paws induced by injection of SNP (0.03 mg/kg) in WT mice pre-treated (red symbols) or not (open circles) with sumatriptan. Data illustrated depict SNP responses 21 days after minipump implantation. **p < 0.01, ***p < 0.001 compared to WT saline, SNP with Mann–Whitney non-parametric test. c Hind paw withdrawal responses of sumatriptan-treated Nav1.9−/− mice compared with sumatriptan-treated WT littermates. *p < 0.05, **p < 0.01, ***p < 0.001 compared to WT with Mann–Whitney non-parametric test. d Comparison of SNP- induced changes in hind paw withdrawal thresholds in sumatriptan-treated WT mice (red symbols), saline-treated Nav1.9−/− mice (open squares) and sumatriptan-treated Nav1.9−/− mice (blue squares). All tests were made at day 21. **p < 0.01, ***p < 0.001 compared to WT sumatriptan with two-way ANOVA followed by Student–Newman–Keuls test. e Normalized periorbital withdrawal threshold plotted as a function of time after SNP injection in saline- treated WT mice (open bars), sumatriptan-treated WT mice (red bars) and sumatriptan-treated Nav1.9−/− mice (blue bars). Data illustrated depict SNP responses 21 days after pump implantation. *p < 0.05, ***p < 0.001; two-way ANOVA followed by Student–Newman–Keuls test sensorimotor gating reflecting the sensitization of mice to weak increased sensitivity of MOH mice to NO. Given the difficulty to sounds30. PPI was measured as the innate reduction of the startle unambiguously isolate Nav1.9 from Nav1.8 currents when using reflex induced by a weak pre-stimulus sound (Fig. 3c). Following CsCl-based pipette solution11,13,14, retrogradely labeled dural SNP injection at day 21, the PPI value was 18.5 ± 2% in saline- afferent neurons were studied from Nav1.8−/− mice. We pro- treated WT mice but reached 42.4 ± 2.5% in sumatriptan-treated vided evidence that inactivation of the Scn10a gene encoding WT mice (Fig. 3c, d), indicating sensitization to sounds. By Nav1.8 did not affect sumatriptan-induced latent sensitization contrast, the PPI value was not significantly different in saline- and hypersensitivity to SNP in MOH mice (Supplementary treated versus sumatriptan-treated Nav1.9−/− mice (Fig. 3c, d). Fig. 3). Together, these data indicate that Nav1.9 is essential to the Nav1.9 current recorded in dural afferent Nav1.8−/− neurons development of NO-induced symptoms of central sensitization cultured at day 21 (i.e. 21 days after pump implantation) was observed in MOH mice. identifiable from its slow activation kinetics and incomplete inactivation, producing ‘persistent’ TTX-resistant Na+ currents. Triptan overuse promotes coupling of NO-cGMP to Nav1.9 Neither the mean Nav1.9 peak current density (Fig. 4b), nor the channels. We investigated the molecular basis that promotes level of Nav1.9 mRNA expression (at day 21) (Supplementary

4 NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3 ARTICLE

a Light sensitivity (day 21) b Auditory threshold (day 21) Saline Saline Sumatriptan Sumatriptan 8 100 Light 25 7 ns ns Sound stimulus (60–120 dB SPL, 10 kHz) 10 20 9 12 12 95 10

15 6 90

Light box Dark box 10 30 min test 85

5 SNP injection (dB SPL) Strain gauge 80 Acoustic startle reflex threshold after

Time in the dark after SNP injection (min) WT Nav1.9–/– WT Nav1.9–/– cd Sound sensitization after SNP injection (day 21) Sound sensitization after SNP injection (day 21) WT mice Nav1.9–/– mice 115 dB 115 dB Saline SPL 70 dB SPL 70 dB Sumatriptan SPL SPL 5 40 Saline Saline ns PPI PPI 30 6 6 10 % a.u. 10 % a.u. 20 7 Sumatriptan Sumatriptan PPI PPI 10 Pre-pulse inhibition (%) 0 –/– 100 ms WT Nav1.9

Fig. 3 Deletion of Nav1.9 prevents NO-induced photophobia and phonophobia in MOH mice. a Light sensitivity of sumatriptan-treated and saline-treated WT and Nav1.9−/− mice 2 h after injection of SNP (0.03 mg/kg) at day 21. Visible light: 380–740 nm, 2600 lx. Behavioral tests were made for a duration of 30 min. ns not significant; **p < 0.01; Mann–Whitney test. b ASR threshold in sumatriptan-treated and saline-treated WT and Nav1.9−/− mice 2 h after injection of SNP at day 21. ns not significant; **p < 0.01; Mann–Whitney test. c SNP-induced pre-pulse inhibition: the PPI was measured using the protocol illustrated (top panels) in saline- (middle panel) and sumatriptan- (bottom panel) WT and Nav1.9−/− mice 2 h after injection of SNP. a.u. arbitrary unit. d Comparison of the PPI in sumatriptan-treated or saline-treated WT and Nav1.9−/− mice 2 h after SNP injection. Tests were made at day 21. ns not significant; **p < 0.01; Mann–Whitney test

Fig. 4) was significantly different in TG neurons from saline- Fig. 6A–C). These data indicate that NO-cGMP pathway activates treated and sumatriptan-treated mice. Nav1.9 in sumatriptan-treated, but not in saline-treated dural SNP (1 mM) increased Nav1.9 peak current (measured at peak afferent neurons. I–V) by 232 ± 11% (n = 20) in dural afferent neurons from − − sumatriptan-treated Nav1.8 / mice, whereas it had little effect Relief of PKA inhibition causes NO coupling to Nav1.9 in (72.7 ± 4%, n = 23) in control dural afferent neurons (Fig. 4a, b). MOH mice. To probe the molecular changes in TG neurons at Current–voltage relationships determined before and after SNP day 21 from sumatriptan-treated mice, we made qPCR analysis of application in sumatriptan-treated dural neurons showed a cGMP-linked signaling molecules. Relative mRNA quantification − 22 mV negative shift in activation V1/2 value (from 11.3 ± 1.8 showed there was a three-fold decrease of the transcript of protein to −32.7 ± 1.02 mV). This shift was also associated with a ~2-fold kinase cAMP-activated catalytic subunit alpha (PKA-Cα) but no increase in Nav1.9 maximum conductance (from −1.28 to −2.64 changes (0.5 < RQ < 2) for the cyclic nucleotide phosphodies- nS/pF) (Fig. 4c, d). By contrast, SNP induced a ~−6 mV shift in terases 3a, 3b and 5a (PDE3a, PDE3b, and PDE5a), the sGC, the dural afferent neurons from saline-treated mice (from −14.47 ± predominant receptor for NO, the Protein Kinase G type I (PKG- 1.3 to −20.8 ± 0.8 mV), which was not associated with change in I) and the Adenylyl Cyclase type III (AC-III) (Supplementary Gmax (Fig. 4d). Fig. 7A). Pre-treatment of 8-Br-cAMP, a membrane permeable Because the cyclic guanosine monophosphate (cGMP) signal cAMP analog, inhibited cGMP-mediated activation of Nav1.9 in pathway plays an important role in NO signaling, we sought to dural afferent neurons from saline-treated mice, although it had determine the involvement of the soluble guanylyl cyclase (sGC) no effect per se on Nav1.9 (Supplementary Fig. 7B–D), indicating in the activation of Nav1.9. Application of methylene blue (100 that cAMP can inhibit cGMP coupling to Nav1.9 in TG neurons. µM), a sGC inhibitor, abolished the effect of SNP on Nav1.9 in + − − Consistently, we found that DiI -dural afferent neurons showed dural afferent neurons from sumatriptan-treated Nav1.8 / mice strong staining for the catalytic PKA subunit phospho T197 (Supplementary Fig. 5). Consistently, the cell-permeable cGMP antibody that recognizes the active form of the PKA protein analog 8-Br-cGMP (1 mM) strongly increased Nav1.9 current (Supplementary Fig. 8A, B). In addition, the mean intensity and negatively shifted V activation by 19 mV (from −13.5 to 1/2 − − (arbitrary unit) per pixel of PKA subunit phospho T197 staining −32.5 mV) in saline-treated Nav1.8 / mice (Supplementary was significantly (p = 0.03, unpaired t-test) reduced from 131 ± 4

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3

a –/– b Saline-treated Nav1.8 mouse (day 21) 15 s Pre-SNP ns SNP –20 mV 200 pA –100 mV 40 15 30 15 20 Ctr SNP (-pA/pF) 10

Current density 0 Saline Suma. Sumatriptan-treated Nav1.8–/– mouse (day 21) SNP –10 mV –100 mV 20 250 Ctr 200 150 SNP 100 300 pA 23

Nav1.9 (%) Nav1.9 50

Increase in peak 0 Saline Suma. 50 ms SNP

cdPre-SNP Saline-treatedPost-SNP Sumatriptan-treated Saline-treated Sumatriptan-treated nS/pF V (mV) V (mV) nS/pF –80 –60 –40 –20 20 –80 –60 –40 –20 20 Pre-SNP 3 Pre-SNP 3 –200 –200 Post-SNP Post-SNP –600 –600 2 2 –1000 –1000

–1400 –1400 1 1 –1800 –1800

I Nav1.9 (pA) I Nav1.9 (pA) –60 –40 –20 0 20 –60 –40 –20 0 20 V (mV) V (mV)

Fig. 4 Sumatriptan treatment promotes activation of Nav1.9 by NO. a Nav1.9 current exposed to 1 mM SNP in dural afferent neurons from saline-treated (top panel) and sumatriptan-treated (bottom panel) Nav1.8−/− mice. CsCl-only-based patch pipette solution. Right-most traces: superimposed Nav1.9 currents before and after SNP application. b Nav1.9 current density (top panel) in dural afferent neurons from saline-treated and sumatriptan-treated Nav1.8−/− mice. ns, not signiﬁcant; unpaired t-test. Bottom panel: mean increase in Nav1.9 peak current induced by SNP (1 mM) in dural afferent neurons from saline-treated and sumatriptan-treated Nav1.8−/− mice. ***p < 0.001; unpaired t-test. c Nav1.9 I–V determined in dural afferent neurons from saline- treated and sumatriptan-treated Nav1.8−/− mice before and after SNP exposure. Insets: superimposed Nav1.9 current traces evoked by voltage steps from −80 to +10 mV from a Vh of −100 mV. Note that not all traces are shown for clarity sake. d Activation curves for Nav1.9 current determined in DiI+-dural neurons before and after SNP application. Boltzmann ﬁts gave V1/2 values of −14.47 ± 1.3 and −20.8 ± 0.8 mV before and after SNP in saline-treated Nav1.8−/− mice (n = 11) and of −11.3 ± 1.8 and −32.7 ± 1.02 mV before and after SNP in sumaptriptan-treated Nav1.8−/− mice (n = 9), respectively. All data collected from neurons cultured at day 21

(n = 44 DiI+-neurons) in control mice injected with saline phasic discharges into multi-action potential (AP) responses in solution to 118 ± 5 (n = 50 DiI+-neurons) in mice treated with 78.5% of dural afferent neurons from sumatriptan-treated WT sumatriptan (data not shown). Changes in PKA expression in mice (Fig. 5a–c). In addition, SNP reduced by 35% the current TGs from sumatriptan-treated mice (n = 8) were also evaluated threshold for firing in neurons from sumatriptan-treated WT by quantifying band intensities on phospho T197 PKA blots and mice (Fig. 5d, e). By contrast, SNP had no significant effects on comparing to blot band intensities in saline-treated mice (n = 8) the firing response and AP current threshold in dural afferent as controls. Densitometry analysis of background-subtracted blots neurons from saline-treated WT mice (Fig. 5c, d). Moreover, SNP from 20 µg of total lysate showed a 22% decrease in phospho caused no changes in excitation or AP current threshold in dural T197 PKA expression in sumatriptan-treated mice versus control afferent neurons from sumatriptan-treated Nav1.9−/− mice mice (Supplementary Fig. 8C). This decrease however did not (Supplementary Fig. 9). reach significant level due to sample variability (Supplementary We tested whether Nav1.9-mediated hyperexcitability regulates Fig. 8D). the secretion of CGRP, a key player in headache pathogenesis Collectively, these data suggest that nitrergic activation of (Fig. 6a). SNP, at low concentrations, had no significant effects on Nav1.9 in sumatriptan-treated mice may result from a decrease in basal secreted CGRP levels in TG cultures from saline-treated PKA activity in dural afferent neurons and subsequent relief of WT mice but enhanced CGRP secretion (+70%) in TG cultures PKA-mediated inhibition of NO–Nav1.9 coupling. from sumatriptan-treated WT mice. Enhanced secretion of CGRP by SNP was not observed in TG cultures from sumatriptan- treated Nav1.9−/− mice (Fig. 6a). Nav1.9-mediated hyperexcitability causes central sensitization. The in vivo consequence of Nav1.9-dependent CGRP secretion Firing activity of retrogradely labeled dural afferent neurons was was tested on SNP-mediated extracephalic mechanical allodynia α studied at day 21. SNP-mediated Nav1.9 activation converted at day 21 using the CGRP antagonist -CGRP8-37. Injection of

6 NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3 ARTICLE

ab100 pA 100 pA 15 s 0 mV Saline-treated WT mouse (day 21) Ctr SNP 500 ms –100 mV (1)

SNP (2) i = 0 pA (1) (2)

40 mV Ctr 100 pA (3) Sumatriptan-treated WT mouse (day 21) SNP SNP (4) i = 0 pA

(3) (4)

100 pA 40 mV

cd eControl Saline Suma. WT mice SNP Saline-treated WT mouse Sumatriptan-treated WT mouse ns 11/14 5 80 6 11 14 Ctr SNP Ctr SNP 4 60 4 3 9 40 5 5/24 2 2 20 1 Responsive cells (%) Responsive 0 increase in firing Fold 0 SNP SNP Δ

± 20 pA 10 ms 40 mV Current threshold (pA/pF) 0 Saline Suma.

Fig. 5 Nav1.9 activation by NO lowers excitability threshold and enhances firing in dural afferent neurons. a Effect of SNP (1 mM) on DiI+ dural afferent neurons from saline-treated (upper panel) and sumatriptan-treated (bottom panel) WT mice. KCl-based intracellular solution throughout. b I–V relationships determined using a slow (50 mV/s) voltage ramp command in DiI+ dural neurons illustrated in a before (1,3) and during (2,4) SNP application. Note the activation of Nav1.9 (inwardly flowing current) by SNP (4). c Percentage of DiI+ neurons responding to SNP (left panel) and mean change in their firing rate (right panel). Protocol as in a.**p < 0.01; Mann–Whitney test. d Generation of APs before and after SNP exposure in dural afferent neurons from saline-treated or sumatriptan-treated WT mice. Steady bias currents were used to maintain the neurons at ~−65 mV. e Comparison of normalized current threshold for AP before and during SNP application in DiI+ dural neurons from saline-treated and sumatriptan-treated WT mice. ns not significant, *p < 0.05 compared to saline-treated WT mice with Wilcoxon matched paired test. All data collected from TG neurons cultured at day 21

α -CGRP8-37 (1 mg/kg), but not the vehicle, reduced SNP-induced mice (Supplementary Fig. 10A) and on SNP-induced allodynia extracephalic allodynia in sumatriptan-treated WT mice, indicat- in sumatriptan-treated Nav1.9−/− mice (Supplementary ing that released CGRP contributes to central sensitization Fig. 10C). α fi (Fig. 6b). Importantly, injection of -CGRP8-37 had no effects on We nally sought to determine whether MC degranulation residual SNP-induced allodynia in sumatriptan-treated Nav1.9−/− contributes to dural afferent terminal excitation through animals (Fig. 6c), providing further evidence that Nav1.9 receptor-driven modulation of Nav1.9. Patch clamp recordings activation by SNP is a prerequisite for CGRP release from showed that the MC mediators histamine and PGE2 are meningeal nociceptors. powerful activators of Nav1.9 current in dural afferent neurons (Fig. 7). Superfusion of histamine and PGE2 increased peak Nav1.9 current by 100.5 ± 25% and 233 ± 40%, respectively, in Nav1.9 mediates vasodilatation and mast cell degranulation. saline-treated Nav1.8 KO mice (Fig. 7a–g). Both mediators Functional consequences of Nav1.9 activation on meningeal caused a substantial leftward shift in the activation curve of microcirculation was examined at day 21 using a laser Doppler Nav1.9 (Fig. 7c, f) and strongly increased the firing rate of dural blood perfusion scanner. SNP (0.03 mg/kg), injected through afferent neurons (Fig. 7h, i). These effects were seen irrespective the jugular vein, caused a gradual increase of meningeal blood of the mouse treatment (Fig. 7g). By contrast, CGRP, SP, and flow in sumatriptan-treated WT mice. No change was seen in neurokinin A, which are potentially released by meningeal saline-treated WT mice and in sumatriptan-treated Nav1.9−/− terminals, had no detectable effects on Nav1.9 current in dural mice (Fig. 6d, e). We further tested whether activation of afferent neurons from either saline-treated or sumatriptan- meningeal nociceptors by SNP causes degranulation of dural treated mice (Supplementary Fig. 11A–D). Thus, some MC mast cells (MCs) and pain amplification through a Nav1.9- mediators have the capacity to activate Nav1.9, resulting in a dependent mechanism. The MC stabilizing agent, sodium feedforward loop that potentiates neurogenic inflammation and cromoglycate (SCG, 10 mg/kg i.p.), injected 30 min prior to nociceptive transmission. Collectively, these results indicate SNP, significantly reduced both the intensity and duration of that Nav1.9 acts as a hub in meningeal nociceptors and the SNP-induced heightened allodynia in sumatriptan-treated contributes to maladaptive nociceptive signal, neurogenic WT mice (Supplementary Fig. 10B) but had no significant inflammation, meningeal vasodilatation, and mast cell degra- effects on SNP-induced basal allodynia in saline-treated WT nulation (Fig. 8).

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3

a bcSumatriptan-treated WT mice Sumatriptan-treated Nav1.9–/– mice TG cultures (day 21) (day 21) (day 21) SNP SNP α -CGRP8–37 α-CGRP8–37 Vehicle or vehicle or vehicle 1.2 1.2 100 2 SNP KCI 2 1.0 1.0 80 Vehicle Vehicle 2 2 α-CGRP8–37 α-CGRP8–37 5 0.8 0.8 ns 60 ns ns 0.6 0.6 4 6 5 4 6 40 4 4 0.4 0.4 SNP SNP

Secretion of CGRP (pg/ml) 20 0.2 0.2 α-CGRP8–37 α-CGRP8–37 Normalized tactile withdrawal threshold Normalized tactile withdrawal threshold Normalized tactile withdrawal 0 0.0 0.0 –101234 –101234 WT, saline WT, suma KO, saline KO, suma Time (h) Time (h)

de–/– Saline-treated WT mouse –/– Saline-treated Nav1.9 mouse WT mice (day 21) Nav1.9 mice (day 21) SNP Pre Post SNP Pre Post Saline Saline 110 Suma 110 Suma

105 105

Sumatriptan-treated WT mouse Suma.-treated Nav1.9–/– mouse 100 100 Pre Post Pre Post

95 Arbitrary unit (%) 95 Arbitrary unit (%) Pre Post Pre Post

90 90 0 20406080 0 20406080 Time (min) Time (min)

Fig. 6 Nav1.9 activation sustains CGRP release, which contributes to central sensitization and meningeal vasodilatation. a Effects of SNP (1 mM), HBSS (vehicle), and KCl (40 mM) on CGRP secretion in TG cultures from WT or Nav1.9−/− mice treated or not with sumatriptan. The n number refers to the number of triplicates. **p < 0.01; Mann–Whitney test. b, c Effect of intravenous injection of α-CGRP8–37 (1 mg/kg) or its vehicle (NaCl 0.9%) on SNP- −/− induced mechanical paw allodynia in sumatriptan-treated WT b or Nav1.9 c mice.**p < 0.01 compared to α-CGRP8–37 pre-injection with Wilcoxon- matched paired test (n = 5 per group). Behavioral tests were carried out at day 21. d–e SNP-induced meningeal blood ﬂow changes in WT d and Nav1.9−/− e mice, treated or not with sumatriptan. *p < 0.05, **p < 0.01; Mann–Whitney test (n = 6 per group). Right panels: representative laser Doppler images taken before and 60 min after SNP injection. The blue color represents low perfusion areas, green and yellow refer to higher perfusion and red shows the highest microperfusion. Scale: 8 mm

Discussion evaluated using multiple headache-like responses in each indivi- How chronic exposure to abortive medication leads to MOH dual animal, including cutaneous facial and extracephalic allo- remains unclear. Our study suggests that chronic use of triptans dynia as well as aversion to light and noise. Our data show that induces MOH through abnormal activation of meningeal Nav1.9 chronic sumatriptan treatment of mice induces a state of dormant by NO. Thus, triptan-overuse headache derives from Nav1.9 sensitization characterized by sharp, exacerbated sensory activation in the TG system, triggering pain facilitation through responses to NO8,9,40. central and peripheral sensitization and inflammation in the Clinical and preclinical studies have consistently demonstrated meninges. increased excitability of the TG system after medication overuse. The complex pathophysiology behind MOH is still only partly TG neuronal hyperexcitability may facilitate the process of per- known. Mechanisms involved may differ from one class of ipheral and central sensitization. We show that NO was capable overused drug to another. Previous studies have shown that of producing strong activation of dural afferent neurons in MOH chronic use of opioids and triptans increases CGRP levels which mice and that sensory hypersensitivity and MOH-associated is well known to be involved in neurogenic inflammation and symptoms were prevented by deleting Nav1.9 but not Nav1.8. headache pain8,31. Impaired diffuse noxious inhibitory controls6 Nav1.9 activation sustains the hyperexcitability of meningeal and central sensitization are also seen in MOH patients4,32. Many nociceptors and lowers the threshold response for afferent pain of these phenomena are similar to mechanisms seen in depen- signaling. Importantly, coupling of NO to Nav1.9 was weak under dence processes33,34. normal conditions, consistent with the observation that deleting Our approach to modeling MOH symptoms was the quanti- Nav1.9 had no impact on SNP-induced (basal) cutaneous allo- fication of increased sensory sensitivity in response to NO, one of dynia in saline-treated mice. Thus, chronic sumatriptan treatment the most common reported trigger for headache and promotes coupling of NO to Nav1.9 channels in dural afferent migraine8,9,35,36. NO donors reliably trigger headache in normal neurons, thus lowering the threshold of the animal’s susceptibility subjects, but trigger migraine and severe pain in migraineurs36, to respond to initiating factors of headache. Importantly, we and this condition is accompanied by an increase in blood levels found that hypersensitivity to SNP was greater in MOH female of CGRP, which is directly linked to the severity of headache than in male mice, which parallels the sexual dimorphism pain37–39. In our study, MOH following NO infusion was reported in MOH and migraine in humans.

8 NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3 ARTICLE

a bc Saline-treated Nav1.8–/– mouse (day 21) –20 mV Pre-histamine 15 s 1.2 –100 mV V (mV) Post-histamine Histamine (100 μM) 1.0 –80 –60 –40 –20 20

0.8 –400 0.6 –1000 G/Gmax 800 pA 0.4 –1600 0.2 –2200 I (pA) –60 –40 –20 0 20 V (mV) def–25 mV –/– 7 s Saline-treated Nav1.8 mouse (day 21) –100 mV PGE2 (500 nM) V (mV) Pre-PGE2 –80 –60 –40 –20 20 Post-PGE2 1.2 1.0

–500 0.8

–1000 250 pA 0.6 G/Gmax –1500 0.4

–2000 0.2 I (pA)

–80 –60 –40 –20 0 20 V (mV) ghSaline-treated WT mouse (day 21) μ Saline Suma. Histamine 100 M

300 6 7 ns l Histamine ns 200 ns 40 mV Saline 5 Suma. 8 11 13 500 ms 16 s 4 100 7 3 2 1 Increase in peak Nav1.9 (%) Increase in peak Nav1.9 0 Hist. PGE2 Hist. PGE2 Fold increase in firing Fold 0

Fig. 7 Mast cell mediators activate Nav1.9 in dural afferent neurons. a Nav1.9 current challenged with 100 µM histamine in a DiI+ dural afferent neuron from a saline-treated Nav1.8−/− mouse. CsCl-only-based patch pipette solution. b I–V relationships from the cell depicted in a determined before and after histamine exposure. c Averaged activation curve for Nav1.9 determined before and after histamine application in DiI+ neurons (n = 13). Single Boltzmann

fits gave V1/2 values of −15 ± 0.3 and −29.3 ± 0.15 mV before and after histamine application, respectively. d Nav1.9 current challenged with 500 nM PGE2 in a DiI+ dural afferent neuron from a saline-treated Nav1.8−/− mouse. CsCl-only-based patch pipette solution. e I–V relationships from the cell depicted in d determined before and after PGE2 exposure. f Averaged activation curve for Nav1.9 determined before and after PGE2 application in DiI+ dural afferent neurons (n = 7). Single Boltzmann fits gave V1/2 values of −17.3 ± 0.6 and −45.6 ± 1.1 mV before and after PGE2 application, respectively. g Increase in Nav1.9 current (at peak I/V as in b and e) induced by histamine in DiI+ neurons from saline-treated or sumatriptan-treated Nav1.8−/− mice. ns not significant, Mann–Whitney test. h Effects of histamine on a DiI+ dural afferent neuron from a WT mouse treated with saline solution. Voltage responses were evoked by depolarizing pulses (+50 pA) applied every 8 s. KCl-based intracellular solution. i Mean increase in firing of dural afferent neurons in response to histamine from WT mice treated with sumatriptan or the saline solution. ns not significant, Mann–Whitney test

What then favors Nav1.9 activation by NO/cGMP in CGRP was prevented by deleting Nav1.9. These data indicate that sumatriptan-treated mice? PKA-Cα transcripts were found to be Nav1.9-dependent release of CGRP, and possibly other neuro- down-regulated in TGs from sumatriptan-treated animals at day peptides including SP, plays a pivotal role in the vasodilatation of 21, suggesting that constitutive activity of PKA, in control ani- meningeal blood vessels. Our data however do not specify whe- mals, inhibits NO–Nav1.9 coupling. This is consistent with the ther the vasodilatation contributes to MOH-related symptoms or observation that cAMP pretreatment inhibited cGMP-mediated whether this is a side-phenomenon42. Therefore, if vasodilatation activation of Nav1.9 in dural afferent neurons. How chronic of meningeal arteries contributes to the propagation of the cas- stimulation of Gi-coupled 5HT1B/D receptors, which are cade of symptoms, it is likely in conjunction with other factors. notably expressed in TG neuron plasma membrane41, leads to CGRP is also poised to enhance headache pain by central long-term changes of PKA gene transcription remains to be mechanisms. Consistently, our results show that SNP-induced determined. extracephalic allodynia was transiently alleviated by acute α Genetic deletion of Nav1.9 prevented SNP-induced meningeal administration of the CGRP antagonist -CGRP8–37 in vasodilatation, indicating that Nav1.9-dependent excitation of TG sumatriptan-treated animals. Because, extracephalic allodynia is a neurons was a prerequisite for this effect. Thus, Nav1.9 activation manifestation of central sensitization43, these data argue that by SNP not only causes hyperexcitability of nociceptors but also Nav1.9-dependent CGRP release may also occurs at postsynaptic triggers the release of vasoactive neuropeptides in the meninges. structures, such as the trigeminal nucleus caudalis, activity of In vitro experiments further indicated that NO-induced release of which may sensitize thalamic neurons. These results call for an

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3

Sterile neurogenic inflammation

+ Meningeal vessel CGRP + relaxation SP + Central sensitization Mast cell degranulation BK 5HT Mechanical allodynia + Triptase + Phonophobia CGRP Nav1.9 + Photophobia + His Feedforward loops + PGE2

NO + Stress Food Odors Latent sensitization Alcohol Hormones Medication overuse

Fig. 8 Central role of Nav1.9 in MOH mechanisms. The following scenario summarizes the contribution of Nav1.9 to MOH. NO, which may be released from different sources, activates Nav1.9 channels in dural afferent neurons from chronically treated mice with triptans. Nav1.9 activation by NO increases the excitability of meningeal nociceptors, which sensitizes central structures leading to extracephalic allodynia, photophobia, and phonophobia. Nav1.9- dependent secretion of CGRP in the meninges, possibly in combination with other mediators, causes degranulation of resident mast cells. By releasing histamine and PGE2, MCs retro-excite meningeal nociceptors through Nav1.9 potentiation. Vasoactive peptides also contribute to vascular relaxation that may further facilitate endothelial (and possibly extravascular) NO production. The consequence is a vicious circle that leads to enhanced activation of meningeal nociceptors and maladaptive pain early use of anti-Nav1.9 drugs that target meningeal nociceptors, littermates. The Scn11a-GAL reporter mouse was previously described15. An IRES/ before the development of central sensitization. LacZpA cassette followed by a loxP/neo/loxP cassette was inserted into the end of Our data demonstrate that Nav1.9-mediated release of neuro- exon 5 of the Scn11a gene. peptides from meningeal nociceptors could trigger degranulation of dural MCs. CGRP and SP, which are often co-released, are Infusion of sumatriptan and injection of SNP. Alzet osmotic minipumps (model particularly important in this regard as they are known to sti- 1007D, Charles River, France) with a nominal flow rate of 0.5 μl/h for 6 days were 44 used for drug infusion. The minipumps were implanted subcutaneously under mulate MC degranulation in different systems . Histamine, a fl 45 anesthesia with iso urane. The day of the pump implant was considered as day 0. major amine released from MCs , and PGE2 another pro- Drugs administered by infusion were sumatriptan (0.6 mg/kg/day, Sigma, St. Louis inflammatory MC mediators46, are well-recognized migraino- USA) and its vehicle (NaCl, 0.9%). SNP was injected subcutaneously into the loose genic substances known to induce pain upon infusion into skin over the neck at 0.03 mg/kg. Infusion of sumatriptan/saline and injection of human subjects diagnosed with migraine. Our data show that SNP/saline were made blind, e.g. the investigator was not aware of the content of the minipump, nor of the nature of the solution (SNP or vehicle) injected on day both substances cause a prominent increase of Nav1.9 and pro- 21. Animals were randomly assigned to treatment groups. mote neuronal hyperexcitability, resulting in a vicious, self- reinforcing cycle of sterile inflammation and nociception (cf Fig. 8). Because MCs express a variety of 5-HT receptors, it is Tactile sensory testing. Hind paw mechanical threshold was assessed using von Frey filaments (Bioseb, France) as described20. For facial testing, mice were sub- possible that chronic treatment with sumatriptan reinforces the jected to a von Frey stimulus applied to the forehead surface, repeated three times possibility that MCs respond to peptide-induced degranulation. (at minimum 30 s interval). The head withdrawal tactile sensory threshold was the However, the reported effects of 5-HT on MCs have generally lowest force to elicit withdrawal in 2 of 3 trials. Data points were normalized to the been found to be mediated by 5-HT1A and not by 5HT1B/D control sensory threshold values determined just before minipump implantation receptors47. (i.e. at day 0) or before SNP injection (i.e. at day 21, H0). In conclusion, our study identifies NO-induced Nav1.9 channel activation as a triggering mechanism for MOH-related symp- Light aversive test. To evaluate the light-aversive behavior after SNP injection, we toms. The way in which medication overuse transforms episodic used a light–dark test. The light-aversion chamber consisted of two equally sized compartments (10 cm width, 13 cm length, and 13 cm height), one painted white migraine into chronic daily headache is still unknown. The pic- and lacking a top, the other painted black and fully enclosed. A corridor (7.2 cm ture that emerges from our study is that abnormal sensitivity of width, 7.2 cm length, and 13 cm height) connects the two compartments. Mice meningeal Nav1.9 channels to the migraine trigger NO may cause were acclimatized at least 1 h in their cages in the testing room and were allowed to the headache phase in MOH patients. Activation of Nav1.9 may explore the light-aversion chamber 10 min before testing. The light-aversive behavior was examined 2 h after SNP injection (0.03 mg/kg) and during 30 min. be a common denominator of overused drugs and migraine The thermal-neutral fiber-optic source is located in the middle of white box and triggers. Therefore, the use of Nav1.9 channel inhibitors, in produces a light intensity of 2600 lx inside the lit area of the white chamber. Mouse combination with sumatriptan or other headache medications, transitions were interpreted to be a reflection of light aversion. may represent a new acute and preventive option for migraine treatment. Acoustic startle threshold and prepulse inhibition tests. The Startle Reflex System (Bioseb, France) was used to measure the acoustic sensitivity of mice. The hardware consisted of an isolation cabinet that minimizes the effects of extraneous Methods noise and vibrations. The startle chamber is situated in the center of the isolation Animals. This project was approved by the Institutional Review Board of the cabinet (250 (W) × 250 (D) × 250(H) mm) and a restrainer (90 × 30 mm) was used regional ethic committee. All animals were used in accordance with the European to restrain minimally the animal during testing. Mice were acclimatized in the Community guiding in the care and use of animals (2010/63/UE). All efforts were restrainer for 5 min per day, one week before the testing session. Mice were also made to minimize the number of animals used and their suffering. Mice (10–12- acclimatized in their cages in the testing room, 2 h before the testing session. week-old adult males and females, C57Bl/6J background) used in this study were Acoustic stimuli for behavioral tests were based on previously published studies Nav1.9−/−, Nav1.8−/− (gifts from N.J. Wood, see refs. 19,23) and their WT defining criteria applicable to mice at around 2 months of age. The waveforms

10 NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3 ARTICLE generated by the mouse’s movement during the startle response were analyzed 100. CGRP antibodies were diluted in PBS containing 5% fish gelatin and applied using the Packwin software (Panlab, Inc., Harvard apparatus). to tissues to be incubated overnight at RT in sealed humidified chambers. Two hours after SNP injection, the mouse was acclimatized for 5 min in the After incubation with the primary antibodies, fixed and fresh frozen tissues presence of a background noise level of 60 dB SPL. The mouse was then exposed to were identically proceeded. Tissues were washed three times for 5 min in PBS and random trials of white noise bursts ranging from 60 to 120 dB SPL at 10 kHz, incubated for 45 min at RT with secondary antibodies diluted in blocking buffer. frequency that falls within the most sensitive region of the mouse audiogram. Each Secondary antibodies were: Alexa Fluor 647-conjugated donkey anti-mouse (1/400, sound intensity was presented three times. The ASR threshold was taken as the Life Technologies), TRITC-conjugated donkey anti-rabbit (1/400, Jackson minimum intensity required to elicit a response in two out of third trials. The PPI ImmunoResearch, Suffolk, UK), TRITC-conjugated donkey anti-rat (1/100, is a reliable, robust quantitative phenotype that is useful for probing sensory Jackson ImmunoResearch), Alexa Fluor 488-conjugated donkey anti-goat (1/200, functions. Testing consisted of a series of 115 dB SPL-test pulses immediately Life Technologies). After six 5 min-washes in PBS, sections were mounted in preceded or not by a prepulse of 70 dB SPL at 10 kHz. PPI was calculated as Mowiol (Sigma-Aldrich). Images were acquired using a LSM 780 laser-scanning follows: PPI (%) = (1–(startle response with prepulse)/(startle response without confocal microscope (Zeiss), initially processed using ZEN software (Zeiss) and prepulse)) × 100. later exported into Adobe Photoshop (Adobe Systems, San Jose, CA) for final processing. Cultures of TG neurons. Mice were anesthetized at day 21 with isoflurane and killed by decapitation. TGs were dissected out and freed from their connective Staining for beta-galactosidase enzyme activity. Evaluation of β-gal enzymatic tissue sheaths. TG neurons were incubated in enzyme solution containing 2 mg/ml activity was performed on transgenic mice that express β-gal at the Scn11a loci. of collagenase IA (Sigma) for 45 min at 37 °C. The tissue was washed several times Once dissected out, the TGs were slightly fixed 15 min in PBS containing 2% PFA, ’ fi and triturated in Hanks balanced salt solution. The resulting suspension was l- 0.2% glutaraldehyde, and 2 mM MgCl2. The tissues were washed with two changes fi tered (70 µm lters) and centrifuged (800×g for 5 min) and plated on poly-L-lysine/ of PBS containing 2 mM MgCl2 for 5 and 8 min. The staining is achieved with an laminin (0.05 and 0.01 mg/ml, respectively) coated Nunclon dishes. Culture overnight incubation of TGs in PBS added with 4 mM Ferrocyanide, 4 mM Fer- ’ fi ’ medium was Dulbecco s modi ed Eagle s medium supplemented with 10% heat- ricyanide, 2 mM MgCl2, 1% Tween-20, and 0.2 mg/ml X-gal at 37 °C. The TGs inactivated FCS, 100 U/ml penicillin–streptomycin, 2 mM L-glutamine, 25 ng/ml were finally post-fixed for 15 min in PBS containing 2% PFA, 0.2% glutaraldehyde, nerve growth factor (NGF), and 2 ng/ml glial-derived neurotrophic factor (GDNF). and 2 mM MgCl2, and washed four times in PBS. The TGs were cryoprotected in PBS containing 4% sucrose for 30 min and then incubated overnight in PBS plus 20% sucrose at 4 °C. The TGs were then frozen in OCT embedding matrix Patch clamp recordings. Patch pipettes had resistances of 2 MΩ. Voltage clamp + (Cellpath, Hemel Hempstead, UK) bathed in chilled isopentane. TGs were then recordings of Na currents used the following intracellular solutions: (CsF-con- sagitally cryosectioned at 14–18 µm, transferred to SuperFrostPlus slides (Fisher taining, in mM), 30 CsF, 100 CsCl, 10 HEPES, 10 EGTA, 8 NaCl, 1 MgCl2, 1 CaCl2, Scientific, Houston, TX), dried at RT and mounted in Mowiol (Sigma-Aldrich). 4 MgATP, 0.4 Na2GTP (pH 7.35, 300 mOsm/l); (CsCl-based, in mM), 130 CsCl, 10 HEPES, 10 EGTA, 8 NaCl, 1 MgCl2, 1 CaCl2, 4 MgATP, 0.4 Na2GTP (pH 7.35, 300 mOsm/l). The extracellular solution contained (in mM): 60 NaCl, 110 sucrose, CGRP staining in TGs from Scn11a-GAL reporter mice. β-gal enzymatic activity – 3 KCl, 1 MgCl2, 10 HEPES, 2.5 CaCl2, 10 glucose, 10 TEA Cl, 0.0005 TTX, in TGs from Scn11a-GAL reporter transgenic mice was revealed as above, except 1 amiloride, 0.05 La3+ (pH 7.4, 305 mOsm/l). For current clamp recording, the that glutaraldehyde was omitted from the fixative solution. Once achieved, TGs fi intracellular solution (KCl-based) consisted of (mM): 115 KCl, 10 HEPES, were post- xed for 2 h in PBS containing 2% PFA and 2 mM MgCl2, washed four 10 EGTA, 8 NaCl, 1 MgCl2, 1 CaCl2, 4 MgATP, 0.4 Na2GTP. The extracellular times in PBS and then cryoprotected in PBS containing 25% sucrose overnight. solution consisted of (in mM) 131 NaCl, 3 KCl, 1 MgCl2, 10 HEPES, 2.5 CaCl2, TGs were then frozen in OCT embedding matrix bathed in chilled isopentane and 10 glucose (pH 7.4, 305 mOsm/l). All chemicals were from Sigma-Aldrich, except sagitally cryosectioned at 14–18 µm, transferred to SuperFrostPlus slides, dried at TTX (Alomone Labs). RT and stored at −80 °C. Slides with TGs were thawed for 15 min at RT and then PCLAMP 9.2 (Axon Instruments Inc.) and PRISM 4.0 (GraphPad) software suites incubated for 1 h 30 min at RT in blocking solution containing 3% BSA and 0.1% were used to perform linear and nonlinear fitting of data. Conductance–voltage curves Tx-100. CGRP antibody (rabbit polyclonal #PC205L, Millipore) were diluted at 1/ = − were calculated from the peak current according to the equation G I/(V Erev), 200 in blocking buffer and applied to tissues to be incubated overnight at 4 °C in fi where V is the test pulse potential and Erev the reversal potential calculated according sealed humidi ed chambers. Tissues were washed three times for 10 min in PBS to the Nernst equation. The activation curve (G−V)wasfitted using the Boltzmann and incubated 40 min at RT in PBS containing 3% BSA added with secondary = + − function: G/Gmax 1/(1 exp[(V1/2 V)/k]), where G/Gmax is the normalized antibodies Alexa Fluor 488-conjugated donkey anti-rabbit (1/200, Life Technolo- conductance, V1/2 is the potential of half-maximum channel activation, and k is the gies). After successive 5 min washes in PBS, sections were mounted in Mowiol steepness factor. (Sigma-Aldrich). Images were acquired using a conventional fluorescence microscope (Zeiss Axio-observer) with constant acquisition settings. Tracer application onto the dura. Mice were anesthetized with isoflurane. Throughout surgery, the core temperature of the mouse was monitored and Quantitation of anti-PKA immunostaining. The TG ganglia were carefully dis- maintained by an homeothermic blanket system for rodents. Two small cranial sected out from NaCl-treated and sumatriptan-treated mice whose dura were DiI- windows were made in parietal bones and the retrograde nerve tracer DiI (DiI stained (see above). The TGs were fixed in 4% PFA in PBS for 2 h 30 min at 4 °C tissue labeling paste, Invitrogen) was applied onto the dura. The bone flaps were then cryoprotected in PBS containing 25% sucrose overnight at 4 °C. The TGs were then replaced after the procedure with bone wax in order to prevent tracer then frozen in OCT embedding matrix bathed in chilled isopentane and stored at spreading. Animals were euthanized for TG extraction 2 days after DiI application. −80 °C until processed. The day before PKA staining, the TGs were sagitally cryosectioned at 14 µm, transferred to SuperFrostPlus slides and stored at −80 °C. Slides with PAF TGs were thawed 15 min at RT and then incubated for 30 min at Tissue preparation, immunostaining, and confocal imaging . For Nav1.9 RT in blocking solution containing 3% fish gelatin and 0.05% saponin. PKA immunostaining, the tissues were cryoprotected in PBS containing 4% sucrose antibodies (rabbit polyclonal #ab75991, AbCam) were diluted at 1/200 in blocking during 30 min and then incubated for at least 1 h in PBS plus 20% sucrose at 4 °C. buffer and applied to tissues to be incubated overnight at 4 °C in sealed humidified The TGs were frozen in OCT embedding matrix bathed in chilled isopentane. The – chambers. Tissues were washed three times for 10 min in PBS and incubated for 3 h TGs were then sagitally cryosectioned at 14 18 µm, transferred to SuperFrostPlus at RT in PBS containing 3% fish gelatin added with secondary antibodies Alexa slides and stored at −80 °C until processed. Whole mount dura maters were − Fluor 488-conjugated donkey anti-rabbit (1/200, Life Technologies). After four transferred to SuperFrostPlus slides, frozen on dry ice and stored at 80 °C until 5 min washes in PBS, sections were stained with DAPI, washed once in PBS and processed. Primary antibodies used and dilutions were as follows: anti-peripherin mounted in Mowiol (Sigma-Aldrich). 1/400 (mouse monoclonal, Millipore, Temecula, CA); anti-NF200 1/400 (chicken In order to compare the fluorescence intensities of PKA immunostaining polyclonal, Aves Labs, Tigard, OR); anti-CD31 (1/400, rat polyclonal, BD Bios- 10 between NaCl-treated and sumatriptan-treated DiI-positive TG neurons, images ciences, Belgium); anti-Nav1.9 L23, (1/100, rabbit polyclonal) ; and anti-CGRP (1/ were acquired using a LSM 780 laser-scanning confocal microscope (Zeiss) using 300, goat polyclonal, AbCam). constant acquisition settings. ImageJ software was used to measure intensities of For CD31 immunostaining, the tissues were fixed for 30 min at room fl fi uorescence on planar projections of confocal raw images spanning 6 µm. Elliptical temperature (RT) in Antigen x (Microm Microtech, France) before being regions of interest encompassing the soma of each DiI-positive neuron were cryoprotected and frozen as described above. Slides with fresh frozen tissues or fi fi fi de ned in the DAPI image to exclude staining of the nucleus. The measurements xed with Antigen x tissues were thawed at RT and then incubated for 90 min at were repeated in an adjacent area out of the tissue and the resulting background RT in blocking solution containing 3% BSA and 0.1% Triton X-100. Primary intensities were subtracted from the soma fluorescence signal. Results are expressed antibodies were diluted in PBS containing 3% BSA and applied to tissues to be fi as mean grey value intensity per pixel meaning the sum of grey values of all pixels incubated overnight at 4 °C in sealed humidi ed chambers. in the ROI divided by the number of pixels. For CGRP immunostaining, the tissues were fixed in 4% PFA in PBS for 3 h at RT then cryoprotected in PBS containing 25% sucrose overnight at 4 °C. The tissues were transferred to SuperFrostPlus slides, frozen and stored at −80 °C until PKA western blot analysis. TGs were collected at 21 days after sumatriptan or processed. Slides with PAF-fixed tissues were thawed at RT and then incubated for saline infusion (T0), scrapped in TE buffer (66 mM Tris pH 6.8, 2% SDS, 10% 1 h30 min at RT in blocking solution containing 5% fish gelatin and 0.2% Triton X- glycerol, 0.1 M DTT, and antiprotease 2×) and sonicated 3 × 5 s. The protein

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3

− Table 1 Primers used in q-PCR analysis were also generated. cDNA was stored at 20 °C before PCR detection. RT pro- ducts were diluted and amplified in 12 µl reactions using Kapa Sybr® Fast qPCR kits (Kapa Biosystems). GAPDH was selected as reference gene for normalization Gene Accession no. Targeted sequence of the qPCR results. qPCR reactions were run in duplicate on a Applied Biosystems (5′−3′) 7500 Fast Real-time PCR thermocycler using MicroAmp® Fast 96-well reaction plates. Thermocycling parameters were 40 cycles of 95 °C for 3 s for denaturation Nav1.7 XM_006499033 f-CCTTGGCCCCATTAAATCTCT and 60 °C for 30 s for annealing and extension. Parameters and reaction conditions r-TGCTCCTATGAGTGCGTTGAC were identical for all sets of primers. To compare transcript levels between saline Nav1.8 XM_006511991 f-TTGACACAACCTCGCTCTATTCC and sumatriptan TG samples, the relative quantification (RQ) method was used. r-ATTTCACCCTGGGTCTTCTCTCA First, the difference (ΔCt) between the cycle threshold (Ct) values of the target Nav1.9 AF118044 f-CCCTTGTGAGTCTCGCTGAC gene and the GAPDH gene was calculated. Then, the difference (ΔΔCt) between r-GGAGTGGCCGATGATCTTAAT the normalized values with or without sumatriptan was calculated: ΔΔCt = Δ −Δ 5HT1B NM_010482.1 f-TCGTTGCCACCCTTCTTCTG Ctsumatriptan Ctsaline. Finally, to determine the ratio of expression levels in sumatriptan samples versus saline samples, we used the RQ formula as follows r-CGTGGTCGGTGTTCACAAAG = −ΔΔCt 5HT1D NM_008309.4 f-GGGTCAATTCATCAAGAACACA RQ 2 . Applied Biosystems 7500 Software v 2.0.6 was used for analysis. Primers used were described in Table 1. r-GCTTGGAAGCTCTGAGGTGT PDE3a NM_018779 f-AATGGGACCACAAGAGAGGG r-TTCACTCTGGGCTTGTGGAT Statistical analysis. All values are shown as mean ± standard error of the mean (SEM) and n represents the number of animals or cells examined. Except for some PDE3b NM_011055 f-AAACGATCGCCTCTTGGTCT behavioral experiments, no statistical methods were used to pre-determine sample r-CCCAGGGTTGCTTCTTCATC sizes but our sample sizes are similar to those reported in previous publications. PDE5a NM_153422 f-GGAAATGGTGGGACCTTCACT Assessment of normality for sample >14 was tested using the r-AAGAACAATACCACAGAATGCCA Kolmogorov–Smirnov or the D’Agostino–Pearson Omnibus K2 normality test. GC NM_021896 f-TGTTCACCTCTGCAGGTCAT Tests for differences between two normally distributed populations were performed r-CCACACAATATGCATCCCCG using two-tailed t-test. Small sample size lacks power to test normality, therefore PKG2 NM_008926.4 f-CGGAAGAGTGGAGCTTGTTA we typically used non-parametric Mann–Whitney test and Wilcoxon’s test to test for differences between two populations in small samples (n ≤ 15). Two-way r-AGCCATCGGCATCCAGAATTA – – AC3 NM_138305 f- ACATGATGCCCACGATGATA (repeated measures) ANOVA followed by Student Newman Keuls multiple comparison procedure was used for experiments with multiple groups and two r-CAGCAGGATGAGCTGGAAG dependent variables. Figure legends specify which test was used for specific PKAcA BC054834.1 f-GCAAAGGCTACAACAAGGC experiments. Significant levels were set at p ≤ 0.05. Analysis used a combination of r-ATGGCAATCCAGTCAGTCG Clampfit 9.2 (Molecular Devices), Origin 7.0 (OriginLab), and PRISM 7.0 GADPH NM_008084 f-GCAAATTCAACGGCACA (GraphPad) softwares. r-CACCAGTAGACTCCACGAC Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article. concentration was analyzed by spectrophotometer (Nanodrop); concentrations were normalized to 800 µg/ml in TE and bromophenol blue (30 mM). Electro- phoresis was performed via SDS–PAGE using bis/tris polyacrylamide gels 4–12% Data availability fi (Thermo Scientific). After blotting, nitrocellulose membranes were stained with The authors declare that all the data supporting the ndings of this study are included in ponceau red. For immunostaining, membranes were washed in TBST, blocked in the article (or in the Supplementary material) and available from the corresponding blocking solution (5% milk in TBST) and incubated with anti-PKA (rabbit author (P.D.). monoclonal, 1:5000, Abcam, ab75991) and anti-GAPDH (mouse monoclonal, 1:5000, Millipore, MAB374) antibodies overnight at 4 °C. The next day, the Received: 4 December 2018 Accepted: 26 August 2019 membranes were washed three times in TBST, incubated with secondary polyclonal-HRP-conjugated goat anti-rabbit (1:2000, Biorad, 1706515) or anti- mouse (1:2000, polyclonal, Biorad, 1706516) antibody for 1 h at RT and washed again three times in TBST. Afterward, immunoreactivity was visualized using ECL Western blotting substrate (Roche) and densitometrically quantified using a Gbox Analyzer (Syngene). Reported mean grey values were determined with ImageJ. References CGRP secretion. TGs were dissected out at day 21 and cultured for 1 DIV. Cells 1. Bigal, M. E., Lipton, R. B. & Stewart, W. F. The epidemiology and impact of were plated in flat bottom plates at a concentration of 15,000 cells per well. The migraine. Curr. Neurol. Neurosci. Rep. 4,98–104 (2004). supernatant of cultured TG neurons was removed for CGRP detection after 30 min 2. Headache Classification Committee of the International Headache Society incubation with KCl (40 mM), SNP (1 mM), or the vehicle (0.9% NaCl). CGRP (IHS). The International Classification of Headache Disorders, 3rd edition detection was made using an enzyme-linked immunoabsorbent assay kit (Spibio, (beta version). Cephalalgia 33, 629–808 (2013). Berlin Pharma, France). Samples were run in triplicate. 3. Noseda, R. & Burstein, R. Migraine pathophysiology: anatomy of the trigeminovascular pathway and associated neurological symptoms, cortical Laser Doppler blood perfusion scanning. Meningeal microcirculation was spreading depression, sensitization, and modulation of pain. Pain 154(Suppl. determined in isoflurane-anesthesized mice with a laser Doppler blood perfusion 1), S44–S53 (2013). scanner (Periscan PIM-III, Perimed) through the closed cranium after removing 4. Ayzenberg, I. et al. Central sensitization of the trigeminal and somatic the covering skin and the periosteum. Light particles that hit moving blood cells nociceptive systems in medication overuse headache mainly involves cerebral undergo a change in wavelength/frequency, while light particles which encounter supraspinal structures. Cephalalgia 26, 1106–1114 (2006). static structures return unchanged. The scattered light that returns to the tissue 5. de Tommaso, M. et al. Abnormal brain processing of cutaneous pain in surface is registered by a photodetector. The perfusion can be calculated since the patients with chronic migraine. Pain 101,25–32 (2003). magnitude and frequency distribution of the Doppler shifted light are directly 6. Perrotta, A. et al. Sensitisation of spinal cord pain processing in medication related to the number and velocity of blood cells, but unrelated to their direction of overuse headache involves supraspinal pain control. Cephalalgia 30, 272–284 movement. Although the measured parameter is flux, this signal is then processed (2010). to extract information about the microcirculatory blood flow. After a control 7. Kristoffersen, E. S. & Lundqvist, C. Medication-overuse headache: a review. J. period (10 min), SNP (0.03 mg/kg) was injected into the jugular vein catheter and Pain. Res 7, 367–378 (2014). the observation period lasted for 90 min. 8. De Felice, M. et al. Triptan-induced enhancement of neuronal nitric oxide synthase in trigeminal ganglion dural afferents underlies increased – Quantitative real-time RT-PCR. Total RNA was extracted from TG tissue using responsiveness to potential migraine triggers. Brain 133, 2475 2488 (2010). the RNeasy®Plus minikit (Qiagen) and assessed on a NanoDrop ND-2000 spec- 9. De Felice, M. et al. Triptan-induced latent sensitization: a possible basis for – trophotometer (Thermoscientific) for concentration (A260) and purity by OD ratios medication overuse headache. Ann. Neurol. 67, 325 337 (2010). (A260/A280, ranging between 2.0 and 2.2). cDNA was synthesized by using 1.2 µg 10. Padilla, F. et al. Expression and localization of the Nav1.9 sodium channel total RNA from each sample with oligo-dT primers and Superscript III® (Invi- in enteric neurons and in trigeminal sensory endings: implication for trogen) in 20 µl reactions according to the manufacturer’s instructions. To assess intestinal reflex function and orofacial pain. Mol. Cell. Neurosci. 35, 138–152 the specificity of these reactions, no reverse transcriptase and no template controls (2007).

12 NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12197-3 ARTICLE

11. Scroggs, R. S. The distribution of low-threshold TTX-resistant Na+ currents 39. Juhasz, G. et al. NO-induced migraine attack: strong increase in plasma in rat trigeminal ganglion cells. Neuroscience 222, 205–214 (2012). calcitonin gene-related peptide (CGRP) concentration and negative 12. Luiz, A. P., Kopach, O., Santana-Varela, S. & Wood, J. N. The role of Nav1.9 correlation with platelet serotonin release. Pain 106, 461–470 (2003). channel in the development of neuropathic orofacial pain associated with 40. Green, A. L. et al. Increased susceptibility to cortical spreading depression in trigeminal neuralgia. Mol. Pain 11, 72 (2015). an animal model of medication-overuse headache. Cephalalgia 34, 594–604 13. Dib-Hajj, S., Black, J. A., Cummins, T. R. & Waxman, S. G. NaN/Nav1.9: a (2013). sodium channel with unique properties. Trends Neurosci. 25, 253–259 (2002). 41. Ma, Q. P., Hill, R. & Sirinathsinghji, D. Colocalization of CGRP with 5-HT1B/ 14. Coste, B., Osorio, N., Padilla, F., Crest, M. & Delmas, P. Gating and 1D receptors and substance P in trigeminal ganglion neurons in rats. Eur. J. modulation of presumptive Nav1.9 channels in enteric and spinal sensory Neurosci. 13, 2099–2104 (2001). neurons. Mol. Cell. Neurosci. 26, 123–134 (2004). 42. Amin, F. M. et al. Magnetic resonance angiography of intracranial and 15. Maingret, F. et al. Inflammatory mediators increase Nav1.9 current and extracranial arteries in patients with spontaneous migraine without aura: a excitability in nociceptors through a coincident detection mechanism. J. Gen. cross-sectional study. Lancet Neurol. 12, 454–461 (2013). Physiol. 131, 211–225 (2008). 43. Woolf, C. J. Central sensitization: implications for the diagnosis and treatment 16. Baker, M. D., Chandra, S. Y., Ding, Y., Waxman, S. G. & Wood, J. N. GTP- of pain. Pain 152,S2–S15 (2011). induced tetrodotoxin-resistant Na+ current regulates excitability in mouse 44. Forsythe, P. & Bienenstock, J. The mast cell-nerve functional unit: a key and rat small diameter sensory neurons. J. Physiol. 548, 373–382 (2003). component of physiologic and pathophysiologic responses. Chem. Immunol. 17. Dib-Hajj, S. D., Black, J. A. & Waxman, S. G. NaV1.9: a sodium channel Allergy 98, 196–221 (2012). linked to human pain. Nat. Rev. Neurosci. 16, 511–519 (2015). 45. Krabbe, A. A. & Olesen, J. Headache provocation by continuous intravenous 18. Ritter, A. M., Martin, W. J. & Thorneloe, K. S. The voltage-gated sodium infusion of histamine. Clinical results and receptor mechanisms. Pain 8, channel Nav1.9 is required for inflammation-based urinary bladder 253–259 (1980). dysfunction. Neurosci. Lett. 452,28–32 (2009). 46. Antonova, M., Wienecke, T., Olesen, J. & Ashina, M. Prostaglandin E(2) 19. Amsalem, M., Poilbout, C., Ferracci, G., Delmas, P. & Padilla, F. Membrane induces immediate migraine-like attack in migraine patients without aura. cholesterol depletion as a trigger of Nav1.9 channel-mediated inflammatory Cephalalgia 32, 822–833 (2012). pain. EMBO J. 37, e97349 (2018). 47. Kushnir-Sukhov, N. M. et al. 5-Hydroxytryptamine induces mast cell 20. Lolignier, S. et al. Nav1.9 channel contributes to mechanical and heat pain adhesion and migration. J. Immunol. 177, 6422–6432 (2006). hypersensitivity induced by subacute and chronic inflammation. PLoS ONE 6, e23083 (2011). 21. Amaya, F. et al. The voltage-gated sodium channel Na(v)1.9 is an effector of Acknowledgements peripheral inflammatory pain hypersensitivity. J. Neurosci. 26, 12852–12860 This work is supported by grants from the CNRS and the Fondation pour la Recherche (2006). Médicale (FRM 2013 DEQ20130326482 to P.D.). We thank M. Mekaouche, A. Fer- 22. Priest, B. T. et al. Contribution of the tetrodotoxin-resistant voltage-gated nandez and Dr J. Wang (INSERM U1051) for technical assistance, and F. Maingret for sodium channel NaV1.9 to sensory transmission and nociceptive behavior. supervising C.B. during her Master. We thank M. Masse for the gift of the CD31 Proc. Natl Acad. Sci. USA 102, 9382–9387 (2005). antibody. C.B. was supported by the French ministry and Fondation pour la Recherche 23. Lolignier, S. et al. The Nav1.9 channel is a key determinant of cold pain Médicale. sensation and cold allodynia. Cell Rep. 11, 1067–1078 (2015). 24. Huang, J. et al. Gain-of-function mutations in sodium channel Na(v)1.9 in Author contributions painful neuropathy. Brain 137, 1627–1642 (2014). C.B. carried out the behavioral, biochemical, and electrophysiological studies and ana- 25. Zhang, X. Y. et al. Gain-of-function mutations in SCN11A cause familial lyzed the data. J.H. performed electrophysiology. N.O. performed immunostaining stu- episodic pain. Am. J. Hum. Genet. 93, 957–966 (2013). dies. V.P. performed qPCR and western blot. J.R. provided help in behavioral auditory 26. Leipold, E. et al. A de novo gain-of-function mutation in SCN11A causes loss tests. A.D. discussed the project. C.B. and P.D. wrote the paper. P.D. supervised the of pain perception. Nat. Genet. 45, 1399–1404 (2013). project. 27. Han, C. et al. Familial gain-of-function Na(v)1.9 mutation in a painful channelopathy. J. Neurol. Neurosurg. Psychiatry 88, 233–240 (2017). 28. Edvinsson, L. & Uddman, R. Neurobiology in primary headaches. Brain Res. Additional information Rev. 48, 438–456 (2005). Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 29. Markovics, A. et al. Pituitary adenylate cyclase-activating polypeptide plays a 019-12197-3. key role in nitroglycerol-induced trigeminovascular activation in mice. Neurobiol. Dis. 45, 633–644 (2012). Competing interests: The authors declare no competing interests. 30. Geyer, M. A., McIlwain, K. L. & Paylor, R. Mouse genetic models for prepulse inhibition: an early review. Mol. Psychiatry 7, 1039–1053 (2002). Reprints and permission information is available online at http://npg.nature.com/ 31. Belanger, S., Ma, W., Chabot, J. G. & Quirion, R. Expression of calcitonin reprintsandpermissions/ gene-related peptide, substance P and protein kinase C in cultured dorsal root ganglion neurons following chronic exposure to mu, delta and kappa opiates. Peer review information Nature Communications thanks Ingo Kurth, and the other, Neuroscience 115, 441–453 (2002). anonymous, reviewer(s) for their contribution to the peer review of this work. Peer 32. Munksgaard, S. B., Bendtsen, L. & Jensen, R. H. Modulation of central reviewer reports are available. sensitisation by detoxification in MOH: results of a 12-month detoxification study. Cephalalgia 33, 444–453 (2013). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 33. Calabresi, P. & Cupini, L. M. Medication-overuse headache: similarities with published maps and institutional affiliations. drug addiction. Trends Pharm. Sci. 26(Feb), 62–68 (2005). 34. Cupini, L. M., Sarchielli, P. & Calabresi, P. Medication overuse headache: – neurobiological, behavioural and therapeutic aspects. Pain 150, 222 224 Open Access This article is licensed under a Creative Commons (2010). Attribution 4.0 International License, which permits use, sharing, 35. Jansen-Olesen, I., Tfelt-Hansen, P. & Olesen, J. Animal migraine models for adaptation, distribution and reproduction in any medium or format, as long as you give drug development: status and future perspectives. CNS Drugs 27, 1049–1068 appropriate credit to the original author(s) and the source, provide a link to the Creative (2013). Commons license, and indicate if changes were made. The images or other third party 36. Olesen, J. The role of nitric oxide (NO) in migraine, tension-type headache material in this article are included in the article’s Creative Commons license, unless and cluster headache. Pharm. Ther. 120, 157–171 (2008). indicated otherwise in a credit line to the material. If material is not included in the 37. Goadsby, P. J., Edvinsson, L. & Ekman, R. Vasoactive peptide release in the ’ extracerebral circulation of humans during migraine headache. Ann. Neurol. article s Creative Commons license and your intended use is not permitted by statutory 28, 183–187 (1990). regulation or exceeds the permitted use, you will need to obtain permission directly from 38. Sarchielli, P., Alberti, A., Codini, M., Floridi, A. & Gallai, V. Nitric oxide the copyright holder. To view a copy of this license, visit http://creativecommons.org/ metabolites, prostaglandins and trigeminal vasoactive peptides in internal licenses/by/4.0/. jugular vein blood during spontaneous migraine attacks. Cephalalgia 20, – 907 918 (2000). © The Author(s) 2019

NATURE COMMUNICATIONS | (2019) 10:4253 | https://doi.org/10.1038/s41467-019-12197-3 | www.nature.com/naturecommunications 13 1

Maladaptive activation of Nav1.9 channels by Nitric Oxide causes Triptan- induced Medication Overuse Headache

Bonnet et al.,

A CD31/Peripherin B Β-Gal CGRP * * * * * * * * * * d * * 55 µm * 20 µm *

C NF200 Nav1.9 Merge

45 µm

D NF200 Nav1.9 Merge

15 µm

Supplementary Figure 1. Nav1.9 is expressed in most CGRP-positive TG neurons and in few meningeal A-type nerve fibers

(A) Immunostaining for peripherin and CD31 in whole mount of mouse dura mater, showing features of apposition of peripherin-positive fibers with meningeal arteries. Images are projections of 17 consecutive optical sections spanning 18 µm.

(B) TG cryosections from a Scn11a-GAL reporter transgenic mouse was labeled for CGRP (green). β-Gal enzymatic activity is revealed by black dots located in the soma of TG neurons. The quasi-totality of examined Scn11a expressing-TG neurons shows CGRP labelling (asterisks, n = 63/64). Inset: immunostaining omitting the anti-CGRP antibody. Scale, 20 µm.

(C) Example of co-localisation of Nav1.9 and NF200 in nerve fibers from whole-mount of mouse dura mater. Images are projections of 15 consecutive optical sections spanning 20 µm. The rightmost panel shows the merged image with co-localization (yellow) in some fibers (arrows).

(D) Example of absence of co-localization of Nav1.9 and NF200 in nerve fibers from whole- mount of mouse dura mater. Images are projections of 15 consecutive optical sections spanning 20 µm. The rightmost panel shows the merged image with no co-localization (green) in some fibers (arrows).

WT NaCl (n=10) A Day 46 after osmotic minipump implantation B KO NaCl (n=8) Infusion of sumatriptan WT SUMA (n=10) 1.2 SNP Saline (n=4) or saline solution KO SUMA (n=11) Sumatriptan (n=4) 1.4 1.0 1.2

0.8 1.0

0.6 0.8

0.6 0.4 * 0.4 ** **** 0.2 **** 0.2 **

* *** Normalized tactile withdrawal tactile threshold withdrawal Normalized Normalized tactile withdrawal threshold tactile withdrawal Normalized 0.0 0.0 0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 18 20 Time (h) Time (days)

C WT NaCl, NaCl (n=4) WT NaCl, SNP (n=6) D WT SUMA, SNP (n=6) KO NaCl, SNP (n=7) WT SUMA, NaCl (n=4) WT SUMA, SNP (n=6) KO SUMA, SNP (n=10) KO SUMA, NaCl (n=6) 1.4 SNP or SNP or NaCl 1.4 NaCl 1.2 1.2

1.0 1.0 ns 0.8 0.8 ns ns 0.6 ## # 0.6 ns 0.4 0.4 * * * * 0.2 ** ** 0.2 ** **

* * * ** ** Normalized withdrawal threshold withdrawal Normalized Normalized withdrawal threshold withdrawal Normalized 0.0 0.0 0 2 4 6 8 10 0 2 4 6 8 10 Time (h) Time (h)

Supplementary Figure 2. Sumatriptan causes long-lasting hypersensitivity to NO in male mice and causes NO-induced generalized allodynia in female mice through Nav1.9-dependent pathways

(A) Changes in hind paw withdrawal mechanical thresholds induced by injection of SNP (0.03 mg/kg) in WT male mice pre-treated with sumatriptan (filled circles) or the saline solution (open circles), 46 days after minipump implantation. **p<0.01 with Mann-Whitney test.

(B) Hind paw withdrawal responses of sumatriptan-treated Nav1.9-/- female mice (n=11) compared with sumatriptan-treated WT female littermates (n=10). Note the strong reduction of mechanical hypersensitivity in sumatriptan-treated Nav1.9-/- female mice compared to WT female littermates. Infusion of saline solution (vehicle, 0.9% NaCl) in both genotypes does not decrease withdrawal thresholds to tactile stimuli applied to the hind paws. Sumatriptan infusion : 0.6 mg/kg/day for 6 days. *p<0.05, **p<0.01, ***p<0.001 compared to KO SUMA with Mann-Whitney non-parametric test.

(C) Comparison of SNP-induced changes in hind paw withdrawal thresholds in sumatriptan- treated WT female mice (red symbols, n=6) and saline-treated WT female mice (blue symbols, n=6). Note that saline injection causes no hypersensitivity in either sumatriptan-treated or saline-treated WT female mice (downward and upward triangles, respectively). All tests were made at day 21. *p<0.05, **p<0.01 compared to WT NaCl, SNP with two-way ANOVA followed by Student-Newman-Keuls test. # p<0.05, ## p<0.01 compared to WT NaCl, NaCl with two-way ANOVA followed by Student-Newman-Keuls test.

(D) Comparison of SNP-induced changes in hind paw withdrawal thresholds in sumatriptan- treated Nav1.9-/- female mice (blue symbols, n=10), saline-treated Nav1.9-/- female mice (orange lozenges, n=7) and sumatriptan-treated WT female mice (red symbols, n= 6, same data as in C). Note the strong reduction of SNP-induced mechanical allodynia in sumatriptan-treated Nav1.9- /- female mice. *p<0.05, **p<0.01 compared to KO SUMA, SNP with two-way ANOVA followed by Student-Newman-Keuls test. ns, not significant compared to KO SUMA, SNP with two-way ANOVA followed by Student-Newman-Keuls test. All tests were made at day 21.

* ns

Supplementary Figure 3. SNP induces heightened tactile allodynia in sumatriptan-treated Nav1.8-/- mice

(A) Infusion of sumatriptan, but not saline solution (NaCl 0.9%), decreases withdrawal thresholds to tactile stimuli applied to the hind paws of Nav1.8-/- mice (n=8) and WT littermates (n=8). *p<0.05, compared to Nav1.8 KO SUMA with Mann-Whitney non-parametric test.

(B) Changes in withdrawal mechanical thresholds induced by SNP injection (0.03 mg/kg) at day 21 in Nav1.8-/- mice and WT littermates pre-treated or not with sumatriptan. n=8 mice per group. Note that Nav1.8 deletion had no significant effects on SNP-induced heightened allodynia in sumatriptan-treated mice. ns, not significant compared to WT SUMA, SNP with Mann-Whitney non-parametric test.

Supplementary Figure 4. SNP-induced heightened tactile allodynia in sumatriptan-treated mice was not associated with changes in transcriptional expression of Nav1.9

Histogram showing the expression of Nav1.7, Nav1.8 and Nav1.9 transcripts relative to GAPDH as assessed by qPCR. TGs were dissected out at day 21 in mice treated with saline solution or sumatriptan. n indicates the number of mice. ns, not significant compared to saline-treated WT mice with Mann-Whitney non-parametric test.

Supplementary Figure 5. Guanylyl cyclase inhibition prevents SNP-induced activation of Nav1.9 in dural afferent neurons from sumatriptan-treated mice

(A) Effects of the superfusion of methylene blue (100 µM), a sGC inhibitor, on SNP (1 mM) activation of Nav1.9 current recorded in a dural afferent neuron from a sumatriptan-treated Nav1.8-/- mouse. Currents were evoked from a holding potential of -100 to -10 mV every 7 s. CsCl-only-based patch pipette solution. TGs were cultured at day 21.

(B) Histogram showing change in mean Nav1.9 current density before and after application of SNP in dural afferent neurons pre-treated or not with methylene blue (100 µM). Recordings were made from sumatriptan-treated Nav1.8-/- TGs cultured at day 21 (n=8). ns, not significant with Wilcoxon matched paired test.

(C) Averaged activation curves for Nav1.9 determined before and after SNP application in dural afferent neurons (n=8) pretreated with methylene blue. Single Boltzmann fits show no significant changes in activation V1/2 values.

Supplementary Figure 6. The cell-permeable cGMP analog 8-Br-cGMP activates Nav1.9 in dural afferent neurons

(A) Effect of the superfusion of 8-Br-cGMP (1 mM) on Nav1.9 current recorded in a dural afferent neuron from a saline-treated Nav1.8-/- mouse. Currents were evoked from a holding potential of -100 to -30 mV. CsCl-only-based patch pipette solution. TGs were cultured at day 21.

(B) Histogram showing the change in mean Nav1.9 current density before and after application of 8-Br-cGMP (1 mM) in dural afferent neurons (n=12) from saline-treated Nav1.8-/- mice. TGs were cultured at day 21. *p<0.05 with Wilcoxon matched paired test.

(C) Averaged activation curves for Nav1.9 determined before and after 8-Br-cGMP application in dural afferent neurons (n=12). 8-Br-cGMP caused a negative shift of ~19 mV in the activation V1/2 value for Nav1.9.

Supplementary Figure 7. PKA downregulation promotes NO coupling to Nav1.9 channels

(A) Transcriptional expression of the receptors 5HT1B and 5HT1D, the cyclic nucleotide phosphodiesterases PDE3a, PDE3b, PDE5a, the sGC, AC-III and the kinases PKG-I and PKA-Cα. The level of the gene of interest was measured in sumatriptan-treated TG neurons relative to the level of gene expression in control (saline) TG neurons (2-∆∆Ct method). TGs were dissected out at day 21 in mice treated with saline solution or sumatriptan. GAPDH was used as house keeping gene.

(B) Effect of pretreatment of 8-Br-cAMP (1 mM) on 8-Br-cGMP-induced activation of Nav1.9 current recorded in a dural afferent neuron from a saline-treated Nav1.8-/- mouse. Currents were evoked from a holding potential of -100 to -20 mV. CsCl-only-based patch pipette solution. TGs were cultured at day 21.

(C) Mean Nav1.9 current density before and after application of 8-Br-cAMP (1 mM) in dural afferent neurons (n=12, saline-treated Nav1.8-/- mice) and before and after application of 8-Br- cGMP (1 mM) in dural afferent neurons (n=12) pre-treated with 8-Br-cAMP (1 mM) for 10 min. TGs were cultured at day 21. *p<0.05 with Wilcoxon matched paired test.

(D) Averaged activation curves for Nav1.9 determined in the presence of 8-Br-cAMP before (open symbols) and after (filled symbols) 8-Br-cGMP application in dural afferent neurons (n=12, Nav1.8-/- mice).

KDa NaCl Suma 40_ PKA

38_ C GAPDH D 60 40 ns

30 40

20 values(AU) 20

grey 10 Grey valuesGrey(AU)

0 Mean 0 NaCl Suma NaCl Suma

Supplementary Figure 8. Dural afferent neurons show high levels of activated PKA

(A-B) DiI-retrogradely labeled TG neurons (red) were stained with T197, an antibody that recognizes the phosphorylated active form of PKA (green). TGs were collected two days after DiI application through a cranial window in the parietal bone. Note, that some DiI-positive, small diameter, TG neurons showed a high level of constitutively activated PKA. Images are projections of 8 consecutive optical sections spanning 9 µm. Rightmost panel: merged images.

(C-D) Individual (C) or mean (D) Western blot densitometry values of blots probed with anti- T197 PKA from sumatriptan- and saline-treated TGs collected at day 21. Single dot represents 2 TGs from a single mouse. Loading, 20 μg/lane. ns, not significant (p = 0. 46, Mann-Whitney test). Inset shows a Western blot probed with anti-T197 PKA and anti-GAPDH (as control). Note that normalization of PKA signal to GAPDH signal shows a 15 ± 2% decrease in T197 PKA expression in SUMA-treated mice (n=4) compared to saline-treated mice (n=4), although this did not reach statistically significant level (Mann-Whitney test).

Supplementary Figure 9. SNP has no excitatory effects in dural afferent neurons from sumatriptan-treated Nav1.9-/- mice

(A) Current-clamp responses of DiI+ dural afferent neurons from Nav1.9-/- mice treated with sumatriptan. Voltage responses were evoked by 500 ms-depolarizing pulses. The horizontal bars indicate the time and duration of SNP application (1 mM). KCl-based intracellular solution. TGs were cultured at day 21.

(B) Current threshold for AP (normalized to the cell membrane capacitance) before and after SNP exposure in sumatriptan-treated Nav1.9-/- dural afferent neurons. ns, not significant with Wilcoxon matched paired test.

Supplementary Figure 10. Inhibition of mast cell degranulation reduces SNP-induced heightened allodynia

(A-C) Effect of intraperitoneal injection of the MC stabilizing agent SCG (10 mg/kg) on the SNP- induced allodynia in saline-treated WT mice (A), sumatriptan-treated WT mice (B), and sumatriptan-treated Nav1.9-/- mice (C). Injection of SCG or the vehicle (NaCl) was achieved 30 min prior to subcutaneous injection of SNP (0.03 mg/kg). Experiments were performed at day 21. ns, not significant, *p<0.05, **p<0.01, ***p<0.001 compared to vehicle-injected mice with Mann-Whitney non-parametric test.

Supplementary Figure 11. CGRP and substance P have no effects on Nav1.9 in dural afferent neurons from saline and sumatriptan-treated mice

(A-C) Lack of effect of sumatriptan (100 µM, A), CGRP (10 µM, B) and SP (10 µM, C) on Nav1.9 current in dural afferent neurons from saline-treated Nav1.8-/- mice. Currents were evoked from a holding potential of -100 to -20 mV every 15 s. CsCl-only-based patch pipette solution. Cell culture made at day 21.

(D) Histogram showing the mean increase in Nav1.9 peak current induced by PGE2 (500 nM, as positive control), CGRP (10 µM), SP (10 µM), neurokinin-A (Neuro-A, 10 µM) and sumatriptan (Suma, 100 µM) in dural afferent neurons from either saline or sumatriptan-treated Nav1.8-/- mice. Cell culture made at day 21.

Reviewers' comments:

Reviewer #1 (Remarks to the Author):

A technically excellent study of the role of Nav1.9 in NO-induced dural nociceptor sensitization in a model of medication overuse headache (MOH) employing well-integrated genetic, biochemical, electrophysiological and behavioral studies. Theur use of triptan-induced MOH makes these studies highly relevant to our understanding of the response of migraine to one of its primary treatment modalities. As our current knowledge of the mechanism of MOH is extremely limited, their results provide an important mechanistic model for the treatment of this common and debilitating pain syndrome. My suggestions mainly relate to ways in which their studies might be better integrated into the relevant clinical literature.

The title probably should indicate that they are studying a model of MOH.

The relative contribution of serotonin and histamine in the rodent mast cell, as triptans act at serotonin receptors.

Use of male mice in a model of a female predominant pain syndrome.

Clinical observation that “MOH does not develop in persons without a history of headache when medication is being used for other conditions, such as inflammatory diseases,” as it relates to the present studies.

PGE2  cAMP  PKA is a class pathway to sensitize nociceptors/

Congenital insensitivity to pain can be induced by loss of NaV1.9 function. Thus, presumably baseline thresholds are elevated in their knockout mice (hence their use of normalized threshold). This should be noted explicitly.

The trigger of headache by NO donors, in patients with migraine has a delayed onset, on the order of hours. I don’t know if this is true for MOH.

Sodium cromoglycate has not been useful for the treatment of migraine.

The spectrum of the light source?

Page 4. (TTX) Na+ currents  (TTX) sensitive Na+ currents

An outstanding contribution that substantially advances our understanding of medicine induced headache.

Reviewer #2 (Remarks to the Author):

The authors are investigating the role of Nav1.9 in medication over-use headache which is an important clinical problem. They develop a model of MOH based on chronic dosing with sumatriptan and suggest a mechanism whereby this results in reduced expression of PKAC-α, this is associated with enhanced coupling between NO and Nav1.9, trigeminal neuron hyper-excitability and behavioural hypersensitivity to mechanical stimuli, photo and phonophobia. The model is definitely interesting, the identification of Nav1.9 as a key hub in MOH would be a significant advance for the field. There are however a number of important weaknesses that need to be addressed: Major points: 1. Gender: All the mice used in the study were male yet as the authors point out migraine/MOH is much commoner in females. Why were female mice not studied? This is a significant deficiency in the study and at least the major finding that Nav1.9 has a key role in MOH (at a behavioral level) should be replicated female mice.

2. Behavioral analysis and blinding: In the behavioral testing (eg. Fig 2). Why is a normalized mechanical withdrawal threshold given, what is it normalized to? The authors refer to a previous publication but this gives absolute values not normalized data. It is preferable in my view to give absolute values or at least a more detailed and transparent explanation (eg. is this because Nav1.9 -/- have different baseline values?). Peri-orbital testing data is only shown at day 21 given that this is the most relevant aspect for MOH/migraine it should also be given at earlier timepoints (as it has been for hindpaw).

Re. blinding in the reporting section the statement is that ‘The behavioral experiments in which animals were treated with sumatriptan or saline solution chronically (with minipumps), and injected at day 21 with SNP or vehicle were made double blind; i.e. the investigator was not aware of the content of the minipump, nor was he aware of the solution (SNP or vehicle) he was injecting on day 21.’ I was confused by this as All behavioral experiments should be performed blind. Was the investigator also blind to genotype? A statement regarding blinding needs to be made in the manuscript (please note that it doesn’t make sense referring to a double blind experiment in animals).

3. Mechanism by which chronic sumatriptan enhances coupling between NO and Nav1.9: Is there direct evidence that chronic sumatriptan treatment enhances NO release and there are also assays available for assessing cGMP? There is a suggestion that the sumatriptan leads to reduced PKAC-α and this results in less inhibition of cGMP signalling. It would be helpful to look at this further at protein level for instance with western blot analysis to back up the q-PCR data and also looking at the activated phospho T197 PKA form which is currently only examined with immunostaining.

4. The level of statistical reporting could be improved throughout: See comments on blinding above. In the methods section some general comments on the statistical tests used are provided. However it is essential that the exact test used for that specific experiment is given in each case (ie in the actual results we don’t know when a t-test, versus one way versus two way ANOVA was used etc). In ‘A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons’. NA is given. The data should always be tested for normality and I presume that this was done? This should be stated in the statistics section. ‘Sample size was typically determined/chosen to make statistical analysis relevant. For electrophysiological experiments, more than 10 cells were typically analyzed, except for some rare technically challenging experiments (n=6). The rationale for why these sample sizes are sufficient is because it is compatible for the chosen statistical tests. Behavioral studies used a minimum of 10 animals in most tests, to minimize variability between animals. As a rule, no conclusions were derived from small n numbers.’ This is not a great justification of sample size. We don’t choose sample size just to suit the statistical test undertaken. The sample size chosen relates to the variance in the data, the effect size that we think is meaningful as well as the significance level (alpha). According to the ARRIVE guidelines there should be a power calculation performed at the outset to justify sample size. I suggest that the authors rethink this section and perform some proper sample size calculations albeit post hoc.

Minor points 1. Supplemental figure 3 it looks to me that the hypersensitivity induced by sumatriptan is less in the Nav1.8-/- versus WT mice (panel A). What is the result of direct statistical comparison of these two groups?

2. I may have missed it but more information needs to be given re mice expressing β-Gal at the SCN11a locus. At least a reference of where/how this line was generated should be given (this applies to all the transgenic lines). The authors state that double staining of CGRP and Nav1.9 could not be undertaken. Good antibodies to β Galactosidase are available meaning that it would be simple to assess both the co-expression of Nav1.9 immunostaining with β Gal and also β Gal and CGRP. 3. In results page 7. ‘Neither the mean Nav1.9 peak current amplitude (Fig 4B), nor the level of Nav1.9 mRNA…’ Do they mean to say ‘Neither the mean Nav1.9 peak current density?’ which is the top graph in the panel? 4. In current clamp analysis of dural afferents in Figure 5. Do any of these neurons develop spontaneous activity in response to SNP? Is there a change in action potential waveform in response to SNP?

Reviewer #3 (Remarks to the Author):

Bonnet et al. report the role of the voltage-gated sodium channel Nav1.9 in medication overuse headache (MOH). Nav1.9 is a channel involved in pain sensation and generates a persistent sodium current to modulate neuronal activity. It´s mutations are associated with different human pain disorders. In mice, the authors show here that chronic use of triptans induces MOH through abnormal activation of meningeal Nav1.9 channels. Bonnet and colleagues administer the 5-HT receptor agonist sumatriptan, which is approved for acute therapy in migraine and in cluster headache, and show that animals chronically treated with the drug display increased responsiveness / abnormal activation of Nav1.9 to nitric oxide (NO). NO is a well-known trigger for headache and migraine. Phenotypic readout for the “migraine-like symptoms” in mice are generalized allodynia, photophobia, and phonophobia. The authors suggest that Nav1.9-mediated release of neuropeptides from meningeal nociceptors involves degranulation of mast cells and causes inflammatory pain.

This is a very interesting paper dealing with a highly relevant topic and has implications for translational approaches in human pain therapy. The findings are new. Methods and experimental approaches are state-of-the-art and I have only few minor suggestions:

Migraine occurs at a 3:1 female to male ratio. In their experiments, the authors report the results from male mice. Can the authors maybe comment whether sex-specific differences have also been observed using female mice? This would be a very interesting point to address, however, repeating all experiments with a cohort of female mice would exceed the aim of the present study.

Can the authors speculate whether the application of currently available sodium-channel blockers in combination with sumatriptan or other headache medications would be beneficial?

Methods: Please provide more details on the generation of the Nav1.8 and Nav1.9 KO-mice used in the study. NCOMMS-18-36489

Point-by-point reply to Reviewers’ comments.

Reviewer #1 (Remarks to the Author):

A technically excellent study of the role of Nav1.9 in NO-induced dural nociceptor sensitization in a model of medication overuse headache (MOH) employing well- integrated genetic, biochemical, electrophysiological and behavioral studies. Theur use of triptan-induced MOH makes these studies highly relevant to our understanding of the response of migraine to one of its primary treatment modalities. As our current knowledge of the mechanism of MOH is extremely limited, their results provide an important mechanistic model for the treatment of this common and debilitating pain syndrome. My suggestions mainly relate to ways in which their studies might be better integrated into the relevant clinical literature.

We thank the referee for his/her very helpful comments.

The title probably should indicate that they are studying a model of MOH.

We refer now to Triptan in the title, which reads “Maladaptive activation of Nav1.9 channels by Nitric Oxide causes Triptan-induced Medication Overuse Headache”.

The relative contribution of serotonin and histamine in the rodent mast cell, as triptans act at serotonin receptors.

This is a good point. 5-HT is implicated in enhancing inflammatory reactions of skin, lung and many other tissues. Mast cells (MCs) are known to express serotonin receptors. However, in mouse bone marrow-derived MCs and human CD34-derived MCs there is no evidence that 5-HT degranulates MCs or modulates IgE-dependent activation but instead 5-HT seems to promote MC adherence to fibronectin and MC migration. 5-HT1A is the principal receptor mediating the effects of 5-HT in these MCs (Kushnir-Sukhov et al., J Immunol 2006; 177:6422-6432). Thus, these data suggest that 5-HT may sustain inflammation by increasing MCs at the site of tissue injury, but not necessarily through degranulation and release of histamine.

Another study dealing with MCs from the guinea pig small intestine and segments of human jejunum also suggested that 5HT effects are mediated via the 5-HT1A receptor. At variance, this study also showed that stimulation of 5-HT1A can degranulate MCs and release histamine (Wand et al., Am J Physiol Gastrointest Liver Physiol 304: G855–G863, 2013). It is also well known that treatment with anti-5-HT1AR diminishes the severity of contact allergy in experimental animals, an effect mediated by mast cells; while an agonist of this same receptor reduces the stress level and relieves pruritus in patients with atopic dermatitis, a disease also involving mast cells.

Sumatriptan is a 5-HT1D/5-HT1B receptor agonist (Razzaque Z, Heald MA, Pickard JD, et al., 1999, Br J Clin Pharmacol. 47: 75–82) that is supposed not to be active on the 5-HT1A receptor. Therefore, although one cannot definitely rule out that sumatriptan interacts with MCs, the current knowledge supports the idea that 5HT action on MCs is mediated by the 5- HT1A receptor.

A sentence has been added in the discussion on page 14. It reads “Because MCs express a variety of 5-HT receptors, it is possible that chronic treatment with sumatriptan reinforces the possibility that MCs respond to peptide-induced degranulation. However, the reported effects of 5-HT on MCs have generally been found to be mediated by 5-HT1A and not by 5HT1B/D receptors (47).”

Use of male mice in a model of a female predominant pain syndrome.

This is an important point, indeed, which has been raised by the 3 referees and the Editor. It is something we had in mind because the male: female ratio is about 1: 3.

We have now completed a new series of experiments in which the effects of chronic infusion of sumatriptan (or NaCl) and injection of SNP (or NaCl) at day 21 were tested on WT (n=20) and Nav1.9 KO (n=19) mouse females. As before, this series of experiments was made blind.

We treated chronically mouse females with sumatriptan (0.6 mg/kg/day, as with males) and assessed quantitatively generalized mechanical allodynia using von Frey filaments. We found that sumatriptan infusion but not saline solution (vehicle, 0.9%), decreased withdrawal thresholds to tactile stimuli applied to the hind paws of WT female mice. Females appeared more sensitive than males as hind paw withdrawal threshold was sensibly lower and sensory threshold was slower to recover. Similar to males, sumatriptan-treated WT females showed enhanced mechanical hypersensitivity at day 21 in response to SNP injection (0.03 mg/kg). This heightened allodynia was fully prevented in Nav1.9 KO female mice.

These data are now included in the result section on page 6 and in supplementary Figure 2B- D; it reads “Because females have increased risk of developing migraine and MOH, we tested whether infusion of sumatriptan for 6 days (0.6 mg/kg/day) produced mechanical hypersensitivity and latent sensitization to NO in WT female mice as observed for the opposite gender. Chronic sumatriptan produced a strong reduction in mechanical withdrawal thresholds of the hind paw relative to saline-treated WT female mice and to sumatriptan- treated Nav1.9-/- female mice (Supplemental Fig. 2B). Injection of SNP (0.03 mg/kg) at day 21 once sensory thresholds had returned to pre-sumatriptan baseline values caused significantly stronger mechanical allodynia in sumatriptan-treated WT female mice compared to saline- treated WT female mice (Supplemental Fig. 2C). SNP-induced heightened mechanical allodynia was absent in sumatriptan-treated Nav1.9-/- female mice (Supplemental Fig. 2D), reaching similar amplitude to that caused by SNP in saline-treated Nav1.9-/- female mice (Supplemental Fig. 2D). This series of experiments shows that Nav1.9, as observed in male mice, had no role in SNP-induced allodynia in saline-treated animals, but contributes to the heightened SNP allodynia in sumatriptan-treated female mice.” Clinical observation that “MOH does not develop in persons without a history of headache when medication is being used for other conditions, such as inflammatory diseases,” as it relates to the present studies.

This notion has been proposed by specialists in the field (Espen Saxhaug Kristoffersen, Christofer Lundqvist). However, we could not find a report in which people with other conditions (arthritis or inflammatory diseases) were asked to take medication such as simple analgesics more than 15 days per month or a combination analgesics more than 10 days per month for 3 months or more (criteria for MOH, cf Headache Classification Committee of the International Headache Society, 2013), and were then evaluated for headache. Having said that, it is true that if pre-existent headache disorder is required to develop MOH, then treating, presumably non-migrainous, mice with sumatriptan may not be a ‘perfect’ model of MOH. This is clearly the limitation of our mouse model, and of most models dealing with migraine/headache, perhaps with the exception of genetic models. However, because a clear connection has been made between headache-specific pain pathways and headache medication effects in generating a more chronic pain, our model at least addresses the mechanisms of chronic headache medication effects, which resemble phenomena seen in dependence/addiction processes.

PGE2 − cAMP − PKA is a class pathway to sensitize nociceptors cAMP/PKA pathway is indeed known to sensitive and modulates the pain pathway. It is well established in DRG neurons but the situation is not that clear in TG neurons. Levy and Strassman (2002; Journal of Physiology (2002), 538.2, pp. 483–493) have shown that the cAMP-PKA cascade is involved in sensitization of dural mechanonociceptors through different mechanisms operating in separate neuronal populations. But they also showed that 34% of meningeal mechanosensitive units were not sensitized by local application to the dura of dibutyryl adenosine 3‚5‚-cyclic monophosphate (dbcAMP), a stable membrane- permeant cAMP analogue. Activation of PKA- and B-Raf-dependent p38 MAPK pathways in mouse TG neurons has also been shown to decrease membrane excitability through the stimulation of A-type K+ channel (Zhao et al., Cell Signal. 2016 Aug;28(8):979-88). Cannabinoid agonists are known to inhibit TRPV1 in trigeminal ganglion neurons though PKA and PKC pathways (Wang et al., Neurol Sci. 2012 Feb;33(1):79-85). So, PKA pathways (in the cell body? in the terminals?) may also have inhibitory effects.

Our data shows that cAMP inhibits the coupling between NO and Nav1.9, so the resulting effect is a decrease in excitability driven by Nav1.9, but this does not exclude other actions of cAMP that may promote excitation. We found that PGE2 has excitatory effects on dural nociceptors through its potentiation of Nav1.9, but we have evidence that PKA does not mediate its effects.

Insensitivity to pain linked to Nav1.9 mutation (L811P, L1302F) in humans is probably due gain-of-function but not loss-of-function properties of the channel (Huang et al., J Clin Invest. 2017 Jun 30; 127(7): 2805–2814). Nav1.9 mutations that evoke small degrees of membrane depolarization cause hyperexcitability and familial episodic pain disorder or painful neuropathy, while Nav1.9 mutations evoking larger membrane depolarizations generate hypoexcitability (by inactivating the spike-generating system) and insensitivity to pain.

In mice, it is well established by different groups (S. Waxman, J.N. Wood; BT Priest, etc..) that Nav1.9 KO mice have quite normal sensory (noxious heat & mechanical stimuli) thresholds. The only acute phenotype of Nav1.9 KO mice is on noxious cold (Lolignier et al., the Nav1.9 channel is a key determinant of cold pain sensation and cold allodynia, Cell Rep. 2015 11(7):1067-78.). Actually, Nav1.9 channels have a low constitutive activity under normal conditions but are particularly active upon inflammation.

The first reason of using normalized threshold is because of within-strain variation of mechanical withdrawal threshold (from 0.4 to >1 g) even in standardized mouse environment. This also helps comparison between genotypes.

The trigger of headache by NO donors, in patients with migraine has a delayed onset, on the order of hours. I don’t know if this is true for MOH.

We think so, but could not find a specific study in the literature. NO can cause immediate headache in most normal people and cause a delayed headache (5-6 hr after exposure) in people with 1) migraine without aura, 2) chronic tension-type headache, infrequent episodic tension-type headache and frequent episodic tension-type headache. People with cluster headache develop a cluster headache attack 1-2 hours after intake.

Sodium cromoglycate has not been useful for the treatment of migraine.

Sodium cromoglycate has been shown to be effective in controlling gastrointestinal symptoms, but is less effective in other body systems. It is possibly not reaching appropriate concentrations in the meninges. It is poorly absorbed through all body surfaces apart from the bronchial mucosa. When administered orally between 0.8% and 1% is absorbed systemically. However, inhaled sodium cromoglycate has been shown to decrease the symptoms of bone pain, fatigue and headache (Edwards AM, Hagberg H. Oral and inhaled sodium cromoglicate in the management of systemic mastocytosis: a case report. J Med Case Rep. 2010;4:193. Published 2010 Jun 26. doi:10.1186/1752-1947-4-193) and in some cases has also been reported to exert a protective effect on hypersensitivity mechanisms as well as the symptoms of migraine (Monro J, Carini C, Brostoff J. Migraine is a food-allergic disease. Lancet. 1984 Sep 29;2(8405):719-21).

The spectrum of the light source?

Visible light: 380 to 740 nanometers (430–770 THz). Now indicated in the corresponding legend.

Page 4. (TTX) Na+ currents  (TTX) sensitive Na+ currents Thanks. Corrected.

An outstanding contribution that substantially advances our understanding of medicine induced headache.

Thank you very much for your support and helpful comments.

Reviewer #2 (Remarks to the Author):

We would like to thank Reviewer#2 for his/her very constructive comments. We have performed additional experiments to address all the points raised by Reviewer#2.

Major points: 1. Gender: All the mice used in the study were male yet as the authors point out migraine/MOH is much commoner in females. Why were female mice not studied? This is a significant deficiency in the study and at least the major finding that Nav1.9 has a key role in MOH (at a behavioral level) should be replicated female mice.

This is an important point, indeed, which has been raised by the 3 referees and the Editors. It is something we had in mind because the male: female ratio is about 1: 3.

We apologize for the ambiguity. We refer to the paper mainly for the use of von Frey filaments. The first reason of using normalized threshold is because of within-strain variation of mechanical withdrawal threshold even in standardized mouse environment. For example, WT male mice have mechanical withdrawal threshold ranging from 0.4 to 1.5 g. Plotting absolute values makes the graphs messy and hard to figure out. Data points were normalized to Day 0 (the measure obtained just before the implantation of the minipump, e.g. Figure 2A,C) or H0 (the measure made just before the injection of SNP, e.g. Figure 2B,D). This is now indicated in the methods Section on page 15. This also helps comparison between genotypes. However, it is important to note that Nav1.9 KO mice have mechanical withdrawal threshold (evaluated using von Frey filaments) not different from WT mice, as Nav1.9 has no apparent role in mechanical withdrawal threshold under normal conditions. This has been published previously by different groups (Amaya et al. 2006. J. Neurosci. 26:12852–12860 ; Priest et al. 2005. Proc. Natl. Acad. Sci. USA. 102:9382–9387; Lolignier et al., Cell Rep. 2015 May 19;11(7):1067-78; etc..). However, recently Hoffmann et al., (2017, Pain. 158:58-67) suggested that deletion of Nav1.9 reduces noxious sensory thresholds.

About peri-orbital testing: these experiments are really difficult in mice. We have failed to measure peri-orbital thresholds every day (or every 2 days) because mice tend to exhibit increased nocifensive behavior when stimulated repetitively days after days with von Frey filaments in the peri-orbital region. They become very agitated as soon as we move forward the filaments, so it was technically not possible to apply the different filaments onto the face.

Our mistake. The investigator was not aware of the content of the minipump, neither of the solution he/she was injecting (SNP or saline solution) on day 21. However, the investigator was aware of the genotype. The text in the reporting section was corrected and a sentence has been added in the method section on page 15.

Chronic sumatriptan treatment has been shown to increase the level of expression of the NO synthase in trigeminal ganglion dural afferents (De Felice et al., 2010, Brain. 133, 2475-2488; cited in the ms), so there is potential for increased NO release, although this has not been directly tested.

Our data show sumatriptan-treated mice have lower levels of PKA transcripts and reduced PKA subunit phospho T197 immunostaining in meningeal TG neurons compared to saline- treated mice (page 9). Further to Reviewer #2 suggestion, we made a new series of experiments in order to evaluate phospho T197 PKA expression in TGs using western blots. Changes in PKA expression in TGs from sumatriptan-treated mice (8 mice, 16 TGs) were evaluated by quantifying band intensities on phospho T197 PKA blots and comparing to blot band intensities in saline-treated mice (8 mice, 16 TGs) as controls. Densitometry analysis of background-subtracted blots from 20 µg of total lysate showed a 22% decrease in phospho T197 PKA expression in TGs from sumatriptan-treated mice versus controls (see new Supplemental Fig. 8C). However, this decrease did not reach significant level (Mann-Whitney test) due to sample variability (see Supplemental Fig. 8C,D). These data are now presented on page 9 and supplemental Figure 8.

We also normalized intensities of phospho T197 PKA blots to GAPDH blots (data not shown). Normalization of mean PKA signals to GAPDH ones shows a 15 ± 2% decrease in T197 PKA expression in SUMA-treated mice (4 mice, 8 TGs) compared to saline-treated mice (4 mice, 8 TGs). This again did not reach statistically significant level (Mann-Whitney test).

Overall, we think it may be difficult to detect a significant decrease in PKA protein level in whole TGs using Western blot, especially if the downregulation affects a subset of TG neurons.

We now mention in the figure legend the statistical test used for each analysis. We clarified the statistics in the method section and in the Reporting Summary. It reads “All values are shown as mean ± standard error of the mean (SEM) and n represents the number of animals or cells examined. Except for some behavioral experiments, no statistical methods were used to pre-determine sample sizes but our sample sizes are similar to those reported in previous publications. Assessment of normality for sample >14 was tested using the Kolmogorov- Smirnov or the D'Agostino-Pearson Omnibus K2 normality test. Tests for differences between two normally distributed populations were performed using two-tailed t-test. Small sample size lacks power to test normality, therefore we typically used non-parametric Mann– Whitney test and Wilcoxon’s test to test for differences between two populations in small samples (n ≤ 15). Two-way (repeated measures) ANOVA followed by Student-Newman-Keuls multiple comparison procedure was used for experiments with multiple groups and two dependent variables. Figure legends specify which test was used for specific experiments. Significant levels were set at p≤ 0.05. Analysis used a combination of Clampfit 9.2 (Molecular Devices), Origin 7.0 (OriginLab) and PRISM 7.0 (GraphPad) softwares.”

The sample size chosen relates to the variance in the data, the effect size that we think is meaningful as well as the significance level (alpha). According to the ARRIVE guidelines There should be a power calculation performed at the outset to justify sample size.

We did not systematically performed calculation to justify sample size. However, in some behavioral experiments we performed sample size calculations using the software G*Power (http://www.gpower.hhu.de/). We found that sample sizes with a minimum of 8-13 animals were necessary in most experiments, hence our typical sample sizes of 10-15 animals in most tests.

There is indeed a significant difference during the ‘recovery’ phase at day 10 and day 12 between Nav1.8 KO mice versus WT mice treated with sumatriptan (Supplemental Figure 3A; *p<0.05, compared to Nav1.8 KO SUMA with Mann-Whitney non-parametric test). We know that sumatriptan has no acute effect on TG Nav1.8 current and that Nav1.8 transcript level remains stable upon chronic sumatriptan treatment. Therefore, it is plausible that Nav1.8 contributes to maintain hyperexcitability of TG neurons, given its role (along with Nav1.7) in the generation/propagation of spikes in nociceptors.

The Scn11a-GAL reporter mouse was made by JSK and published in 2007 (ref 19). An IRES/LacZpA cassette followed by a loxP/neo/loxP cassette was inserted into the end of exon 5 of the Scn11a gene. Reference to previously published papers is now indicated in the method section. The Nav1.9 and Nav1.8 KO transgenic lines are gifts from Prof N. J. Wood and were previously published (ref 19, 23). We added the references in the method section.

The authors state that double staining of CGRP and Nav1.9 could not be undertaken. Good antibodies to β Galactosidase are available meaning that it would be simple to assess both the co-expression of Nav1.9 immunostaining with β Gal and also β Gal and CGRP.

This is a good suggestion. Unfortunately the two anti-β-Galactosidase polyclonal antibodies we tested on TGs showed strong background level. So, we decided instead to perform immunostaining of CGRP in TGs from Scn11a-GAL reporter mice, and then to reveal β-gal enzymatic activity. This required to remove glutaraldehyde from the fixative solution and once revealed to post-fix the tissue for 2h with 2% PFA. This approach is now described in the method section on page 20. We found that the quasi-totality of CGRP-positive TG neurons exhibited β-gal staining in cryosections from Scn11a-GAL reporter transgenic mice (n = 63/64). These data are now presented on page 5 and Supplemental Figure 1B; they indicate/confirm that CGRP is expressed in Nav1.9-positive TG neurons.

3. In results page 7. ‘Neither the mean Nav1.9 peak current amplitude (Fig 4B), nor the level of Nav1.9 mRNA…’ Do they mean to say ‘Neither the mean Nav1.9 peak current density?’ which is the top graph in the panel?

Yes, we meant ‘current density’. This has been corrected on page 8 of the revised ms.

4. In current clamp analysis of dural afferents in Figure 5. Do any of these neurons develop spontaneous activity in response to SNP? Is there a change in action potential waveform in response to SNP?

We did not see changes in AP waveform. Spontaneous firing in response to SNP application was also a very rare event. SNP occasionally caused small depolarization, but did not generate firing unless the membrane potential was near the negative slope conductance (regenerative process) of the Nav1.9 current. SNP-induced small depolarization may be linked to TRPA1/V1 activation.

Reviewer #3 (Remarks to the Author):

We would like to thank Reviewer #3 for his/her very helpful and positive comments.

This is an important point, indeed, which has been raised by the 3 referees and the Editors. It is something we had in mind because the male: female ratio is about 1: 3.

Can the authors speculate whether the application of currently available sodium-channel blockers in combination with sumatriptan or other headache medications would be beneficial? Yes?

It is an important point indeed and a strategy we are currently developing. As usual, the difficulty resides in the specificity of the inhibitor. We amended the text on page 14 to read: “Therefore, the use of Nav1.9 channel inhibitors, in combination with sumatriptan or other headache medications, may represent a new acute and preventive option for migraine treatment”.

Methods: Please provide more details on the generation of the Nav1.8 and Nav1.9 KO- mice used in the study.

We now provided this information. Reference to previously published papers using these mice is now indicated in the method section. The Scn11a-GAL reporter mouse was made by JSK and published in 2007 (ref 19). An IRES/LacZpA cassette followed by a loxP/neo/loxP cassette was inserted into the end of exon 5 of the Scn11a gene. The Nav1.9 and Nav1.8 KO transgenic lines are gift from Prof N. J. Wood and were previously published (ref 19, 23).

REVIEWERS' COMMENTS:

Reviewer #1 (Remarks to the Author):

The authors have provided new data and well reasoned discussion of the points raised in the intial review. I have but just one suggestion. While the authors nicely executed parallel experiments in female mice, as suggested by all 3 reviewers, they do not npte in the added text that the effect in females was greater than in males, which nicely parallels the sexual dimorphism found in migraine in humans.

Reviewer #2 (Remarks to the Author):

The authors have done an excellent job of revising the manuscript and addressing my concerns I particularly commend the addition of a female cohort of mice.

Reviewer #3 (Remarks to the Author):

All points have been sufficiently addressed. NCOMMS-18-36489A Point-by-point reply to Reviewers’ comments.

We thank the referees for his/her very helpful comments.

Reviewer #1 (Remarks to the Author):

The authors have provided new data and well reasoned discussion of the points raised in the initial review. I have but just one suggestion. While the authors nicely executed parallel experiments in female mice, as suggested by all 3 reviewers, they do not note in the added text that the effect in females was greater than in males, which nicely parallels the sexual dimorphism found in migraine in humans.

We added a comment on page 13 (discussion), it reads: “Importantly, we found that hypersensitivity to SNP was greater in MOH female than in male mice, which parallels the sexual dimorphism reported in MOH and migraine in humans.”

Reviewer #2 (Remarks to the Author):

The authors have done an excellent job of revising the manuscript and addressing my concerns I particularly commend the addition of a female cohort of mice.

Reviewer #3 (Remarks to the Author):

All points have been sufficiently addressed.