Meningeal lymphatic CGRP signaling governs pain via cerebrospinal fluid efflux and neuroinflammation in migraine models

Abstract

Recently developed antimigraine therapeutics targeting calcitonin gene-related peptide (CGRP) signaling are effective, though their sites of activity remain elusive. Notably, the lymphatic vasculature is responsive to CGRP signaling, but whether meningeal lymphatic vessels (MLVs) contribute to migraine pathophysiology is unknown. Mice with lymphatic vasculature deficient in the CGRP receptor (CalcrliLEC mice) treated with nitroglycerin-mediated (NTG-mediated) chronic migraine exhibit reduced pain and light avoidance compared with NTG-treated littermate controls. Gene expression profiles of lymphatic endothelial cells (LECs) isolated from the meninges of Rpl22HA/+;Lyve1Cre RiboTag mice treated with NTG revealed increased MLV-immune interactions compared with cells from untreated mice. Interestingly, the relative abundance of mucosal vascular addressin cell adhesion molecule 1-interacting (MAdCAM1-interacting) CD4+ T cells was increased in the deep cervical lymph nodes of NTG-treated control mice but not in NTG-treated CalcrliLEC mice. Treatment of cultured hLECs with CGRP peptide in vitro induced vascular endothelial-cadherin (VE-cadherin) rearrangement and reduced functional permeability. Likewise, intra cisterna magna injection of CGRP caused rearrangement of VE-cadherin, decreased MLV uptake of cerebrospinal fluid (CSF), and impaired CSF drainage in control mice but not in CalcrliLEC mice. Collectively, these findings reveal a previously unrecognized role for lymphatics in chronic migraine, whereby CGRP signaling primes MLV-immune interactions and reduces CSF efflux.

Keywords: Cell biology; G protein–coupled receptors; Lymph; Pain; Vascular biology.

Figure 1. Calcrl^iLEC mice treated with NTG exhibit partially ameliorated chronic migraine pain.

(A) Experimental protocol representation. (B) Images of mouse facial expression of pain on Day 6, minute 42 following injection. Unit scores are depicted. E, ears; Y, eyes; N, nose; C, cheek. Sum of scores tallied bottom right. (C) Facial expression of pain measured 30 minutes after NTG injection and recorded for 20 minutes. Facial expression of pain is scored once per minute for the 20-minute recording duration and averaged. Mice were allowed to acclimate to the chamber 4 days before first injection (day 1) and 2 days before baseline measurement (day –1). **P < 0.01 between Calcrl^fl/fl and Calcrl^iLEC treated with NTG. Mean grimace score on (D) day –1 (preinjection baseline) grouped by genotype. Colors indicate injection given later in experimental protocol, matching panel B. Mean Grimace Score on (E) day 3, and (F) day 9 of chronic migraine model. For C–F, n = 7–10 animals per group representing 4 independent cohorts. Significance calculated using 2-way ANOVA with Tukey’s multiple comparisons test. Graphs show mean ± SD. (G) Facial expression of pain for Ramp1^–/– animals and controls. ***P < 0.001 between Ramp1^–/– and WT treated with NTG. Mean grimace score on (H) day –1 (preinjection baseline) grouped by genotype. Colors indicate injection given later in experimental protocol, matching key in panel G. Not all animals were recorded for baseline. Mean grimace score on (I) day 1, and (J) day 7 of chronic migraine model. For G–J, n = 8–11 animals per group representing 3 independent cohorts. Significance calculated using 2-way ANOVA with Tukey’s multiple comparisons test. Graphs show mean ± SD.

Figure 2. Calcrl^iLEC mice treated with NTG exhibit ameliorated light avoidance and anxiety behavior.

(A) Schematic of light aversion and movement assay. (B) Schematic of general (top) and daily (bottom) experimental protocol. Mice were allowed to acclimate to the chamber 4 days before the first injection and 2 days before the baseline measurement (baseline measurement recorded on day –1) Light avoidance and movement behavior are measured 30 minutes after NTG injection and were recorded for 30 minutes. (C) Time spent in dark chamber reported for all test days and on day 1 (D) and day 5 (E). Dashed line shows the total test time (1,800 seconds or 30 minutes). (F) Time spent in light chamber reported for all test days. (G) Normalized percent time spent moving in the dark chamber and (H) normalized percent time spent in moving in the light chamber for all experimental days. Percent of time spent moving in the respective chamber were moving normalized to baseline movement in the same chamber data. Data are normalized to the baseline measurement on day –1 by individual calculated as percent time moving on test day divided by baseline percent time moving (day –1) multiplied by 100. *P < 0.05, **P < 0.01 reported between Calcrl^fl/fl and Calcrl^iLEC treated with NTG. Dashed line, baseline. C–H, n = 14–20 animals per group. Data represents 12 independent cohorts. Significance for all data calculated using 2-way ANOVA with Tukey’s multiple comparisons test. P values indicated if < 0.1. Graphs = mean ± SD.

Figure 3. Mice treated with NTG model of chronic migraine exhibit unique MLV translational profiles.

(A) RiboTag Lyve1-Cre schematic depicting experimental protocol. (B) Mean Grimace Score for RiboTag mice, confirming initiation of chronic migraine model. n = 3 animals per group. Data not recorded from final injection day (Day 8). Graph shows mean ± SD. (C) Heatmap of differentially translated genes. Red, up with NTG; Blue, down with NTG. n = 3 animals per group from 1 cohort. (D) IPA analysis of differentially expressed genes from microarray. Top 10 pathways with more than 4 identified transcripts per pathway displayed. (E) Volcano plot of significant differentially genes in MLVs. Labeled genes were selected because PubMed searches resulted in over 7 papers with the topic trend of lymphatic biology. Red text, upregulated and investigated. Significance determined by 1-way ANOVA. Significant P values < 0.05 (above dashed line).

Figure 4. CGRP induces protein level changes in LECs.

(A) qPCR analysis of GJC2 in LECs. (B) Confocal microscopy of connexin-47 (green) and VE-cadherin (magenta) in vehicle (Veh) and CGRP-treated LECs in vitro. Right, Image of overlapping Connexin-47 and VE-cadherin signal. Black represents Connexin-47 and VE-cadherin colocalization. Pixels have signal only if there is signal for both VE-cadherin and Cx-47. Arrows, connexin-47 at continuous VE-cadherin borders. (C) Quantification of MFI of connexin-47. (D) qPCR analysis of PTX3 in CGRP-treated LECs. (E) Immunofluorescence of Pentraxin3 in LECs. (F) Quantification of MFI of Pentraxin3 in LECs. (G) qPCR analysis of MADCAM1. (H) Immunofluorescence of MADCAM1 in CGRP-treated LECs. (I) Quantification of MFI of MADCAM1 in LECs. For all qPCR analysis (A, D, and G), n = 3 biological replicates and with 3 technical replicates. For all immunofluorescence experiments (C, F, and I), n = 3 biological replicates with 3 randomly selected fields of view averaged for each biological replicate. Significance for all data presented calculated using 2-tailed, unpaired student’s t test. Scale bar: 20 μm. Graphs show mean ± SD.

Figure 5. RiboTag and in vitro changes are recapitulated in NTG-mediated chronic migraine.

(A) Whole-mount microscopy of decalcified meninges from mice treated with vehicle or NTG. Secondary-only immunofluorescence included as negative control. Top, costaining of LYVE-1 (green) and Pentraxin3 (magenta). Bottom, increased magnification images of the white dashed square in top row. Black, Pentraxin3. Black dashed lines, MLV outlines. Scale bar: 20 μm. (B) Quantification of PTX3 fluorescence relative to background fluorescence in MLV endpoints. Significance calculated using 1-way ANOVA with n = 3–6 animals with at least 2 endpoints assessed per animal. 2 independent cohorts were assessed. (C) Flow cytometry gating strategy. Quantification of flow cytometric analysis of LPAM1⁺ (α4/β7 integrin⁺) CD4⁺ T cells in (D) DCLNs (draining meninges) and (E) inguinal lymph nodes (distal lymph nodes) of NTG-treated chronic migraine Calcrl^iLEC mice. n = 3 animals per group, pooled left and right DCLN from 2 independent cohorts, performed in duplicate. (F) Schematic indicating proposed relationship between CGRP, MADCAM1, and α4/β7 integrin⁺ (LPAM1⁺) CD4⁺ T cell interaction with LECs. Significance for all graphs calculated using 2-way ANOVA with Tukey’s multiple comparisons test. P value shown if less than 0.05. Graphs show mean ± SD.

Figure 6. CGRP induces formation of continuous, nonpermeable VE-cadherin LEC junctions in cultured hLECs.

(A) LECs treated with media, 100 nM adrenomedullin (AM) (low-permeability control), or 100 nM CGRP and treated with or without CGRP receptor antagonist olcegepant and incubated with antibodies targeting VE-cadherin. White arrows indicate continuous VE-cadherin arrangement, gray arrows indicate discontinuous VE-cadherin arrangement. Scale bar: 10 μm (B) Quantification of proportion of LEC continuous junctions treated with media, CGRP, or adrenomedullin. Significance calculated using 1-way ANOVA with Tukey’s multiple comparison’s test. n = 4–11 biological replicates with 2–3 randomly selected fields quantified per coverslip from 5 independent assays. All visible adherens junctions were counted and scored as either continuous or discontinuous by a blinded scorer. Percent continuous junctions is calculated as number of continuous VE-cadherin junctions divided by total number of observed junctions multiplied by 100. (C) Fluorescence microscopy of LECs grown on biotinylated-fibronectin coated coverslip and treated with 10 μM histamine (high permeability control), 100 nM adrenomedullin, or 100 nM CGRP or media alone then treated with Alexa Fluor-488–labeled streptavidin. Dark signal indicates increased Alexa Fluor-488 streptavidin permeability between LECs. White arrows indicate low permeability cell borders, gray arrows indicate highly permeable cell borders. Scale bar: 20 μm. (D) Permeability between LECs quantified as MFI of bound, labeled streptavidin. n = 5–6 biological replicates with 3–4 randomly selected fields quantified from 3 independent assays. MFI of each field assessed is averaged together to represent 1 biological replicate. Significance for B and D calculated using 1-way ANOVA with Tukey’s multiple comparisons test. Graphs = mean ± SD. (E) Representative Western blot assessing LEC intracellular signaling response to CGRP over 60 minutes. Quantification of phosphor-ERK (p-ERK) to total ERK (t-ERK) (F) p-AKT to t-AKT (G), and p-CREB to t-CREB (H) response to CGRP over 60 minutes. (F–H) significance was determined using an ordinary 1-way ANOVA with Dunnett’s multiple comparisons. 3 independent experiments were conducted.

Figure 7. NTG induced CGRP stimulus is required for formation of continuous VE-cadherin junctions in vivo.

(A) Whole-mount immunofluorescence microscopy targeting LYVE1-1 and VE-cadherin in NTG induced chronic migraine in Calcrl^iLEC mice. White arrows, continuous VE-cadherin junctions, gray arrows, discontinuous VE-cadherin junctions. Scale bar: 10 μm. Black and white inset images are increased magnification of white dashed square. (B) Quantification of proportion of MLV endpoint linear VE-cadherin junctions in NTG-treated chronic migraine Calcrl^iLEC mice. All visible VE-cadherin–positive adherens junctions were counted and scored as either continuous or discontinuous by a blinded scorer. Percent continuous junctions is calculated as number of continuous VE-cadherin junctions divided by total number of observed junctions multiplied by 100. n = 3–4 animals with percent continuous junctions scored from 2–5 meningeal LV endpoints assessed per animal from 2 independent cohorts. (C) Quantification of proportion of MLV endpoint linear VE-cadherin junctions in NTG-treated chronic migraine Ramp1^–/– mice, calculated as in B. n = 4–5 animals per group scored from 2–5 meningeal lymphatic endpoints per animal from 3 independent cohorts. Graph shows mean ± SD. Significance for all graphs calculated using 2-way ANOVA with Tukey’s multiple comparisons test. P value shown if less than 0.05.

Figure 8. CGRP reduces CSF uptake into MLVs.

(A) Schematic of MLV anatomy. Blue dashed square indicates location assessed, while the white dashed square indicates the location for dye measurement, as visualized in B. The asterisk (*) indicates the space outside the sinus. Scale bar: 200 μm. (B) Left, representative image of LYVE1+ (Green) MLVs outlined with white dashed line. Right, EB fluorescence in black. Black dashed line, MLV border. Scale bar: 50 μm. (C) Representative integrated fluorescence intensity across MLV measured using plot profile feature in ImageJ. An average background fluorescence intensity for each field was measured and subtracted from raw EB fluorescence intensity values. Green line, LYVE1 MFI, Blue line, EB MFI, Blue box, margins of MLV. (D) Quantification of background adjusted integral of EB intensity across MLV. n = 3 animals 1 drainage site assessed per animal from 3 independent cohorts. Graph shows mean ± SD. Significance calculated using 2-way ANOVA with Tukey’s multiple comparisons test. P value shown if less than 0.05.

Figure 9. ICM injection of CGRP reduces CSF efflux to the DCLNs by EB dye transport.

(A) Photograph and whole mount fluorescence microscopy of DCLN from Calcrl^iLEC mice injected intra cisterna magna (ICM) with 1 μg/μL CGRP or vehicle diluted in 1% EB. Scale bar: 200 μm. CLV, cervical LV, DCLN, deep cervical lymph node. (B) Quantification of MFI of DCLN of Calcrl^iLEC mice injected ICM with 1 μg/μL CGRP or vehicle diluted in 1% EB. n = 5–7 animals, average MFI of left and right DCLNs graphed from 4 independent cohorts. Graph shows mean ± SD. Significance calculated using 2-way ANOVA with Tukey’s multiple comparisons test. P value shown if less than 0.05. (C) Whole mount fluorescence microscopy of DCLN from Ramp1^–/– or WT mice injected intra cisterna magna (ICM) with 1 μg/μL CGRP or vehicle diluted in 1% EB. Scale bar: 200 μm. (D) Quantification of MFI of DCLN of Ramp1^–/– or WT mice injected ICM with 1 μg/μL CGRP or vehicle diluted in 1% EB. n = 3–4 animals, average MFI of left and right DCLNs graphed from 3 independent cohorts. Graph shows mean ± SD. Significance for calculated using 2-way ANOVA with Tukey’s multiple comparisons test. P value shown if less than 0.05. (E) Representative schematic of CGRP impact on MLVs demonstrating reduced MLV permeability and CSF efflux to the DCLN.