Microglia regulate motor neuron plasticity via reciprocal fractalkine/adenosine signaling

Abstract

Microglia are innate CNS immune cells that play key roles in supporting key CNS functions including brain plasticity. We now report a previously unknown role for microglia in regulating neuroplasticity within spinal phrenic motor neurons, the neurons driving diaphragm contractions and breathing. We demonstrate that microglia regulate phrenic long-term facilitation (pLTF), a form of respiratory memory lasting hours after repetitive exposures to brief periods of low oxygen (acute intermittent hypoxia; AIH) via neuronal/microglial fractalkine signaling. AIH-induced pLTF is regulated by the balance between competing intracellular signaling cascades initiated by serotonin vs adenosine, respectively. Although brainstem raphe neurons release the relevant serotonin, the cellular source of adenosine is unknown. We tested a model in which hypoxia initiates fractalkine signaling between phrenic motor neurons and nearby microglia that triggers extracellular adenosine accumulation. With moderate AIH, phrenic motor neuron adenosine 2A receptor activation undermines serotonin-dominant pLTF; in contrast, severe AIH drives pLTF by a unique, adenosine-dominant mechanism. Phrenic motor neuron fractalkine knockdown, cervical spinal fractalkine receptor inhibition on nearby microglia, and microglial depletion enhance serotonin-dominant pLTF with moderate AIH but suppress adenosine-dominant pLTF with severe AIH. Thus, microglia play novel functions in the healthy spinal cord, regulating hypoxia-induced neuroplasticity within the motor neurons responsible for breathing.

Figure 1.. Fkn increases extracellular adenosine levels in the cervical spinal cord and elicits adenosine-dependent phrenic motor facilitation.

(A): Schematic of experimental setup with intrathecal drug delivery and placement of adenosine (ADO) and inosine (INO) micro-biosensors. (B) Intrathecal fractalkine (Fkn; 100 ng; 12 μL) delivery evoked a slow increase in extracellular adenosine concentration over 30 min (n = 3 independent recordings; Linear Regression, p < 0.001; r = 0.993, r² = 0.985, Adjusted r² = 0.985). Concurrently, phrenic nerve activity was recorded in urethane anesthetized rats maintained at baseline conditions during and after intrathecal Fkn injection (90 min post-delivery). Ventilator volumes were set for each rat based on body mass (0.007 ml * body weight, g; 72–74 breaths per minute). Inspired CO₂ or ventilator frequency were adjusted to maintain end-tidal P_CO2 between 38 and 41 mmHg. Blood gas measurements were taken 2–3 times during the initial baseline, and at 30, 60, and 90-min post-drug (Supplementary Table 1). (C) Schematic of inter-cellular signaling highlighting where receptors/enzymes were blocked/inhibited in D-H. Representative compressed neurograms of integrated phrenic nerve activity are shown for rats that received (D) vehicle (VEH; time controls), (E) VEH + Fkn, (F) CX3CR1 inhibitor, AZD8797 + Fkn, (G) ATPase inhibitor, ARL67156 + Fkn, and (H) A_2A Receptor inhibitor, MSX-3 + Fkn. (I) Phrenic burst amplitude (percent change from baseline; % baseline) was significantly increased 90 min post-Fkn administration (VEH + Fkn); however, CX3CR1, ATPase and A_2A receptor inhibition (schematized in C) attenuated or prevented phrenic motor facilitation (n=6–7 recordings per group; F(4,29) = 21.378, p < 0.001; one-way ANOVA). ****p < 0.001, significant differences vs all groups; Tukey post-hoc Test. Bars show mean ± SEM.

Figure 2.. Severe (vs moderate) hypoxic episodes evoke greater spinal adenosine accumulation.

(A) Adenosine/inosine probes placed between ventral C3/C4 to measure changes in adenosine accumulation during hypoxia. (B) Average traces of extracellular adenosine concentration (µM) during 5 min of moderate (Pa_O2 = 42.7 ± 2.0 mmHg) or severe (Pa_O2 = 27.2 ± 0.8 mmHg) hypoxia (n = 5 per group from 3 rats). Greater adenosine accumulation was observed in severe hypoxic episodes when expressed as (C) peak adenosine level ([ADO]peak; t(8) = −5.299, p < 0.001; unpaired t-test) or (D) total area under the curve ([ADO]AUC; t(8) = −4.218, p = 0.003; unpaired t-test). (E) Pa_O2 strongly correlates with measured extracellular adenosine levels (F(1,14)=206.099, p < 0.0001; r = 0.9677, r² = 0.9364; Adjusted r² = 0.9318 nonlinear regression). Bars are means ± SEM.

Figure 3.. Intrathecal fractalkine differentially regulates moderate (serotonin-dominant) vs severe (adenosine-dominant) AIH-induced pLTF.

Schematic of hypothesized mechanisms for intrathecal and hypoxia-evoked fractalkine (Fkn) release on moderate (A: serotonin-driven Q pathway) vs severe AIH-induced pLTF (B: adenosine-driven S pathway). Phrenic nerve activity was recorded in urethane anesthetized rats during baseline, during intrathecal drug administration, and for 90 min post-treatment while baseline conditions were maintained. Inspired CO₂ and/or ventilator frequency was adjusted to maintain end-tidal P_CO2 between 38 and 41 mmHg. Blood gas measurements were taken 2–3 times during the initial baseline, during the last minute of the first hypoxic episode, and at 30, 60, and 90 min post-AIH (Supplementary Tables 1 and 2). Raw integrated phrenic nerve amplitude at baseline and during maximal chemoreceptor stimulation (10% O₂, 7% CO₂, balance N₂) delivered at the end of each experiment are included in Supplementary Table 3 to assess recording quality. (C, D) Representative compressed neurograms of integrated phrenic nerve activity from rats that received vehicle (VEH; top row) or fractalkine (Fkn; bottom row) ~30 min prior to moderate (C) or severe (D) AIH. Immediately below each neurogram are individual, integrated (∫) phrenic nerve bursts taken during baseline and 90 min post-AIH. One-minute averages of phrenic nerve amplitude were measured at 90 min post-AIH, and are presented as percent change from the pre-AIH baseline value (E); there was a statistically significant interaction between drug (VEH vs Fkn) and AIH protocol (moderate vs severe) on pLTF (n=5–8 independent recordings each group; F(1,23) = 22.316, p < 0.001; two-way ANOVA). *p < 0.001; Tukey post-hoc Test. Bars are means ± SEM.

Figure 4.. Microglia differentially regulate moderate (serotonin-driven) vs severe (adenosine-driven) AIH-induced pLTF.

(A, B) Representative individual, integrated (∫) and raw phrenic (Phr) nerve bursts during baseline and 90 min post-AIH from a vehicle (VEH) control (top row), spinal CX3CR1-inhibited (CX3CR1i) rat, and PLX3397-treated rat (bottom row) prior to moderate (mAIH; A) or severe AIH (sAIH; B). One-minute averages of phrenic nerve burst amplitude were measured 90 min post-AIH, and are presented as percent change from baseline (C). There was a statistically significant interaction between drug (VEH vs CX3CR1i or PLX3397) and AIH protocol (moderate vs severe) on pLTF (n=4–7 independent recordings each group; F(2,32) = 69.896, p < 0.001; two-way ANOVA). *p ≤ 0.001, significant differences vs VEH controls; ‡p < 0.001, significant differences vs sAIH group pretreated with the same drug; Tukey post-hoc Test. (D) Schematic outlining region of the ventral horn where microglia were counted after VEH or PLEX3397 treatment. Representative confocal microscope images from VEH control (E) and PLX3397-treated rats (F) stained for Iba-1 positive microglia (magenta) and phrenic motor neurons (CtB; green). Scale bar (left; 10x magnification): 150 µm; scale bar (right; 40x magnification): 50 µm. (G) Ventral horn Iba-1 positive microglia were counted using a custom code; Iba-1 positive microglia counts were significantly reduced in spinal cords of PLX3397-treated rats versus VEH controls (spinal tissue from n=5–6 rats each group with at least 10 sections per rat; t(9) = 8.347, p = 0.00002; unpaired t-test). Bars are means ± SEM.

Figure 5.. Phrenic motor neuron fractalkine undermines moderate (serotonin-dominant) AIH-induced pLTF but is required for severe (adenosine-dominant) AIH-induced pLTF.

(A) Schematic depicting intrapleural injections for retrograde siRNA transport (nontargeting controls, siNTg; rat CX3CL1/fractalkine, siFkn) and CtB to phrenic motor neurons. (B, C) Representative individual, integrated (∫) and raw phrenic (Phr) nerve bursts taken during baseline and 90 min post-AIH from siNTg-(top row) and siFkn-injected rats (bottom row) prior to moderate (mAIH; B) or severe AIH (sAIH; C). One-minute averages of phrenic nerve amplitude were measured 90 min post-AIH, and are presented as percent change from baseline (D); there was a statistically significant interaction between siRNA (siNTg versus siFkn) and AIH protocol (moderate vs severe) on pLTF (n=4–7 independent recordings per group; F(1,18) = 45.431, p < 0.001; two-way ANOVA). *p < 0.01; Tukey post-hoc Test. (E) Confocal microscope image (10x magnification) of cervical ventral horn; phrenic motor nucleus circled; phrenic motor neurons identified with CtB (green); scale bar (10x magnification): 150 µm. (F) Representative confocal microscope images from siNTg (left) and siFkn (right)-injected groups stained for CtB (green) and Fkn mRNA (red); scale bar (40x magnification): 20 µm. (G) Fkn mRNA fluorescence intensity was significantly reduced in siFkn vs siNTg phrenic motor neurons (spinal tissue from n=5–6 rats per group with at least 10 sections per rat; t(8) = 4.812, p=0.001; unpaired t-test). (H) Fractalkine mRNA fluorescent intensity in non-phrenic motor neurons was similar between siNTg and siFkn groups (t(8) = 0.542, p=0.602; unpaired t-test). Bars are means ± SEM.

Figure 6.. Intercellular model of reciprocal phrenic motor neuron Fkn to microglial CX3CR1 interactions during moderate and severe AIH-induced phrenic motor plasticity.

Hypoxia triggers phrenic motor neuron Fkn signaling in a hypoxia dose-dependent manner. Fkn binds to its receptor, CX3CR1, on nearby microglia, triggering microglia-dependent formation and accumulation of extracellular adenosine (ADO). ADO activates phrenic motor neuron adenosine 2A (A_2A) receptors, constraining the serotonin-dominant (Q) pathway to pLTF during moderate hypoxia (left). In contrast, severe hypoxia, and greater extracellular adenosine accumulation, shifts the serotonin/adenosine balance sufficiently to drive the adenosine-dominant S pathway to pLTF (right).