COX-2-Derived Prostaglandin E2 Produced by Pyramidal Neurons Contributes to Neurovascular Coupling in the Rodent Cerebral Cortex

Abstract

Vasodilatory prostaglandins play a key role in neurovascular coupling (NVC), the tight link between neuronal activity and local cerebral blood flow, but their precise identity, cellular origin and the receptors involved remain unclear. Here we show in rats that NMDA-induced vasodilation and hemodynamic responses evoked by whisker stimulation involve cyclooxygenase-2 (COX-2) activity and activation of the prostaglandin E2 (PgE2) receptors EP2 and EP4. Using liquid chromatography-electrospray ionization-tandem mass spectrometry, we demonstrate that PgE2 is released by NMDA in cortical slices. The characterization of PgE2 producing cells by immunohistochemistry and single-cell reverse transcriptase-PCR revealed that pyramidal cells and not astrocytes are the main cell type equipped for PgE2 synthesis, one third expressing COX-2 systematically associated with a PgE2 synthase. Consistent with their central role in NVC, in vivo optogenetic stimulation of pyramidal cells evoked COX-2-dependent hyperemic responses in mice. These observations identify PgE2 as the main prostaglandin mediating sensory-evoked NVC, pyramidal cells as their principal source and vasodilatory EP2 and EP4 receptors as their targets.

Significance statement: Brain function critically depends on a permanent spatiotemporal match between neuronal activity and blood supply, known as NVC. In the cerebral cortex, prostaglandins are major contributors to NVC. However, their biochemical identity remains elusive and their cellular origins are still under debate. Although astrocytes can induce vasodilations through the release of prostaglandins, the recruitment of this pathway during sensory stimulation is questioned. Using multidisciplinary approaches from single-cell reverse transcriptase-PCR, mass spectrometry, to ex vivo and in vivo pharmacology and optogenetics, we provide compelling evidence identifying PgE2 as the main prostaglandin in NVC, pyramidal neurons as their main cellular source and the vasodilatory EP2 and EP4 receptors as their main targets. These original findings will certainly change the current view of NVC.

Keywords: astrocytes; cerebral cortex; cyclooxygenase-2; interneurons; pyramidal cells.

Figure 1.

Characterization of NMDA-induced vasodilations. A, Diameter changes induced by 5 min bath applications of NMDA (30 μm; black bar and vertical gray zone). The SEM envelopes the mean traces. The horizontal dashed lines represent the initial diameters. The vertical scale bar indicates the relative diameter changes. Under control conditions (Ctrl, black, n = 16 arterioles from 12 rats) NMDA induces vasodilations. Nonselective COX-1/2 inhibition by indomethacin (Indo; magenta, n = 6 arterioles from 3 rats) dramatically reduced the vascular responses. COX-1 inhibition by SC 560 (purple, n = 12 arterioles from 9 rats) did not alter the vascular response in contrast to the selective inhibition of COX-2 by NS 398 (red, n = 10 arterioles from 8 rats). Antagonism of EP2, EP4, and IP receptors respectively by AH 6809 (orange, n = 14 arterioles from 13 rats), L 161–982 (green, n = 11 arterioles from 6 rats) and CAY 10441 (blue, n = 12 arterioles from 7 rats) reduced the vascular responses. Blockade of action potentials by TTX (brown, n = 8 arterioles from 3 rats) dramatically reduced the vascular responses. B, Representative example showing infrared images of a preconstricted diving arteriole that reversibly dilated to NMDA application. Pial surface is upward. Scale bar, 20 μm. White dashed vertical lines indicate the initial position of the vessel wall. C, D, Effects of COX-1/2 inhibition, EP2, EP4, and IP receptor antagonism, and action potentials blockade on maximal amplitude (C) and magnitude (D) of NMDA-induced vasodilations. Dots represent individual arteriolar responses. Bars and error bars correspond to the mean ± SEM. *, **, and *** statistically different from Ctrl condition with p < 0.05, 0.01 and 0.001, respectively; n.s., not statistically significant.

Figure 2.

Vasodilatory prostaglandins release dynamics induced by NMDA. Dynamics of PgE₂ (A) and 6-keto PgF_1α (B) levels in the supernatants (left) and in the slices after treatments (right). Samples were collected before adding COX inhibitors (t = −30), before (t = 0 min) and during (t = 2 and 5 min) NMDA (30 μm; black bar and vertical gray zone), and after removing NMDA (t = 10, 20 and 30 min). PgE₂ (A) and to a lesser extent and 6-keto PgF_1α (B) levels increased during NMDA application (NMDA, black traces) but not after NMDA removal. No change was observed when applying NMDA vehicle (water, Ctrl, gray traces) or when NMDA was applied in presence of NS 398 (red traces) or SC 560 (purple traces). Tissue content (right) in PgE₂ (A) and 6-keto PgF_1α (B) were elevated when slices were treated with NMDA in absence of COX inhibitor. Numbers in parenthesis represent the number of replicates. Error bars are SEM.

Figure 3.

Effects of EP2, EP4, and IP receptors blockade on the NVC response to sensory stimulation. A, Whisker stimulation (gray zone, top black bar) induced increases in CBF in the contralateral barrel cortex under control conditions (Ctrl; black lines). The evoked response was not altered by vehicle (gray line), whereas the EP1/EP2 receptor antagonist AH 6809 (orange, n = 6 rats) and EP4 receptor antagonist L 161–982 (green, n = 4 rats), but not the IP receptor antagonist CAY 10441 (blue, n = 5 rats), decreased this response. The SEM envelopes the mean traces. B, C, Maximal whisker-evoked changes in CBF (B) and in LFPs (C), under baseline (CBF: 8.0 ± 1.1%, 10.9 ± 1.3%, and 8.6 ± 1.2%; LFPs: 100.0 ± 16.7%, 100.0 ± 27.0%, and 100.0 ± 20.8%) and vehicle (CBF: 8.7 ± 1.0%, 11.9 ± 1.2% and 9.0 ± 1.5%; LFPs: 122.3 ± 30.8%, 105.0 ± 14.6%, and 115.4 ± 31.3%) conditions and application of AH 6809 (CBF: 5.9 ± 1.3%; LFPs: 126.6 ± 30.8, n = 6 and 4 rats, respectively), L 161–982 (CBF: 8.7 ± 2.1%; LFPs: 101.9 ± 17.0%, n = 4 rats) or CAY 10441 (CBF: 7.9 ± 1.1%; LFPs: 135.0 ± 38.4%, n = 5 and 3 rats, respectively). Whisker stimulation induced increases in the amplitude of the LFPs in the contralateral barrel cortex as shown by the representative tracings (average of 7 trials from 1 rat) for each compound (C, top). Contralateral LFP amplitudes were normalized by the mean amplitude under baseline conditions. *p < 0.05 and **p < 0.01 compared with vehicle conditions, ★p < 0.05 compared with Ctrl conditions, n.s., not statistically significant. Dots represent individual responses. Horizontal bars and error bars represent mean ± SEM.

Figure 4.

Expression of COX-2 in the neuro-glio-vascular unit. Representative single plane confocal images of double fluorescence staining showing the constitutive expression of COX-2 (red). Scale bar, 100 μm. *Denotes diving blood vessels. Dashed lines represent layer I–II border. Pial surface is upward. A, COX-2 immunolabeling is absent from the vascular bed stained with LEA (green) and (B) from microglia immunostained for Iba-1. C, S100β immunostained astrocytes (green) are essentially COX-2-negative. D, COX-2-immunopositive cells are Satb2-positive (green).

Figure 5.

Sensitivity of RT-mPCR protocols. Total cortical RNAs (500 pg) were subjected to RT-PCR protocols designed for the molecular characterization of perivascular astrocytes (A) and cortical neurons (B). The PCR products were resolved in separate lanes by agarose gel electrophoresis in parallel with Φx174 digested by HaeIII as molecular weight marker and stained with ethidium bromide. The amplified fragments corresponding to astrocytic (A, left) and neuronal (B, top) molecular markers, as well as PgE₂ and PgI₂ synthesizing enzymes (A, right, B, bottom) had the size predicted by their mRNA sequence (Table 2).

Figure 6.

Characterization of prostaglandin producing cells. A, Vital staining of a perivascular astrocyte with SR101. Wide-field fluorescence image of the SR101-labeled cortical astrocyte (left) showing an intensely labeled cell body (arrow) sending a process (arrowhead) onto a diving arteriole (*). Corresponding field-of-view observed under IR-DGC illumination (middle) and superimposition of the two images (right). Pial surface is upward. Scale bar, 10 μm. B, Electrophysiological characterization in voltage-clamp mode of the astrocyte shown in A. Note the linear I/V curve, the low input resistance (slope) and the hyperpolarized resting membrane potential (0 nA intercept) characteristic of passive astrocytes. The inset illustrates current responses evoked by voltage steps (from −180 to 40 mV, 20 mV increments) used to determine the I/V curve at steady-state. C, scRT-PCR analysis of the same astrocyte (A, B) revealing expression of GFAP, S100β, COX-1, and cPGES. D, Voltage responses induced by injection of current pulses in a layer II–III pyramidal cell (bottom traces). Inset, IR-DGC image of the recorded cell, pial surface is upward (scale bar, 10 μm). In response to just-above-threshold current pulse, this neuron fired action potentials with a long lasting biphasic AHP and little frequency adaptation (middle trace). Near saturation, it showed the typical firing of a regular spiking neuron with marked frequency adaptation and spike amplitude accommodation (upper trace). E, The pyramidal cell shown in D expressed vGluT1, CCK, COX-1, COX-2, and cPGES. F, Voltage responses induced by injection of current pulses in a fusiform interneuron (bottom traces). Inset, IR-DGC image of the recorded interneuron, pial surface is upward (scale bar, 10 μm). Note the high input resistance of the cell observed at hyperpolarizing current pulses (middle traces). In response to just-above-threshold current pulse, this interneuron fired a first action potential with a monophasic AHP followed by action potentials with complex AHP. Near saturation, this neuron showed the firing of an adapting-VIP interneuron with marked frequency adaptation and spike amplitude accommodation (upper trace). G, This interneuron expressed vGluT1, GAD65, GAD67, CR, VIP, CCK, COX-2, and cPGES.

Figure 7.

Expression profiles of the PgE₂ and PgI₂ synthesizing enzymes. A, Histograms depicting the occurrence of prostanoid synthesizing enzymes in identified cell types. PGES denotes expression of mPGES1, 2, and/or cPGES. Layer II–III cortical pyramidal neurons (n = 24 cells, black) express much more frequently COX-2, than perivascular astrocytes (n = 44 cells, red) and interneurons (n = 66 cells, white). B, Occurrence of the three PGES isoforms in identified cell types. mPGES1 is not detected in any cell types. mPGES2 is more frequently observed in pyramidal neurons and interneurons than in astrocytes. cPGES is detected in the three cell types but more frequently in pyramidal cells than in astrocytes. *, **, and *** statistically significant with p < 0.05, p <0.01, and p <0.001 respectively. C, Coexpression of prostanoid synthesizing enzymes in identified cell types. The occurrence of cells positive for an enzyme is proportional to the size of the box in a given cell type. Coexpression of PGES (green) and PGIS (blue) with COX-1 (top, purple) and with COX-2 (bottom, red) in astrocytes (S100β and/or GFAP-positive; left), pyramidal cells (vGluT1-positive; middle) and interneurons (GAD65- and/or 67-positive; right).

Figure 8.

Optogenetic stimulations of pyramidal cells evoke COX-2-dependent hyperemic responses. A, Variance filtered speckle contrast image. Darker tones represent higher velocity blood flow. Montage of fractional change in speckle contrast images from cortical optogenetic stimulation of an Emx1-Cre;Ai32 mouse. B, Optogenetically evoked increases in blood flow, the scaled inverse square of speckle contrast, before (black trace) and after application of NS 398 (red trace; n = 5 mice). Blue light excitation (473 nm) in Thy1-YFP mice (yellow; n = 4 mice) does note evoke increase in blood flow. The SEM envelopes the mean traces. C, D, Effects of photostimulation on magnitude (C) and maximal changes (D) of blood flow in Thy1-YFP and Emx1-Cre;Ai32 mice before and after NS 398 application. Dots represent individual neurovascular responses. Horizontal bars and error bars correspond to the mean ± SEM. *Statistically different from Emx1-Cre;Ai32 mice under baseline condition with p < 0.05.

Figure 9.

Biosynthesis and vasodilatory effects of PgE₂ and PgI₂. Increased intracellular calcium activates phospholipase A2 (PLA₂). Arachidonic acid (AA) is released from membrane phospholipids (MPL) by PLA₂ and metabolized by COX-1 or COX-2 to produce the intermediary prostaglandin H₂ (PgH₂). PgE₂ and PgI₂ are, respectively, synthesized from PgH₂ by PGES and PGIS. Upon binding and action on type 2 or 4 PgE₂ (EP2/4) or PgI₂ (IP) receptors these two prostaglandins relax smooth muscle cells (SMCs) or pericytes and dilate blood vessels. AA can be also metabolized by a cytochrome P450-epoxygenase (Cyp450) to produce 14,15-epoxyeicosatrienoic acid (14,15 EET) which induces EP2-dependent vasodilation. EC, Endothelial cell. SC 560 and NS 398 are selective COX-1 and COX-2 inhibitors respectively. Indomethacin (Indo) is a nonselective COX-1/2 inhibitor. AH 6809, L161-982, and CAY 10441 are EP2, EP4, and IP antagonists, respectively.