Ectopic vesicular neurotransmitter release along sensory axons mediates neurovascular coupling via glial calcium signaling

Abstract

Neurotransmitter release generally is considered to occur at active zones of synapses, and ectopic release of neurotransmitters has been demonstrated in a few instances. However, the mechanism of ectopic neurotransmitter release is poorly understood. We took advantage of the intimate morphological and functional proximity of olfactory receptor axons and specialized glial cells, olfactory ensheathing cells (OECs), to study ectopic neurotransmitter release. Axonal stimulation evoked purinergic and glutamatergic Ca²⁺ responses in OECs, indicating ATP and glutamate release. In axons expressing synapto-pHluorin, stimulation evoked an increase in synapto-pHluorin fluorescence, indicative of vesicle fusion. Transmitter release was dependent on Ca²⁺ and could be inhibited by bafilomycin A1 and botulinum toxin A. Ca²⁺ transients in OECs evoked by ATP, axonal stimulation, and laser photolysis of NP-EGTA resulted in constriction of adjacent blood vessels. Our results indicate that ATP and glutamate are released ectopically by vesicles along axons and mediate neurovascular coupling via glial Ca²⁺ signaling.

In the conventional view, a neuron is divided into morphologically and functionally distinct compartments, comprising dendrites receiving synaptic input, axons conducting information via action potentials, and synaptic terminals transmitting information by neurotransmitter release. Recent evidence, however, indicates that neurotransmitters also can be released along axons (e.g., in optic nerve, olfactory nerve and corpus callosum) (1–4). Axonal neurotransmitter release may include the cotransmitter ATP, which is coreleased with glutamate from axons in optic and olfactory nerves (1, 2). However, the mechanisms of this ectopic neurotransmitter release, and in particular the mechanism of neuronal ATP release, are controversial. Early studies suggested vesicular release of ATP at varicosities of peripheral nerves such as the vagus and enteric nerves (5). In few instances, vesicular ATP release could be demonstrated at central synapses (6, 7), and a vesicular ATP transporter has been described recently (8). However, other studies question the hypothesis of vesicular ATP release. Release of ATP from peripheral nerves, for example, is affected differently by presynaptic modulation than is the release of vesicular noradrenaline (9), and suppressing vesicular release with botulinum toxin and tetanus toxin does not inhibit ATP release from cholinergic synaptosomes (10, 11). Alternative modes of ATP release include pores formed by gap junctional hemichannels or P2X7 purinoceptors (12–14). Recent data demonstrate axonal ATP release from cultured neurons through volume-regulated anion channels (VRAC) upon electrical stimulation (15). To date, it is unknown which mechanisms mediate ectopic ATP release in axon tracts such as optic and olfactory nerves.

The main targets of axonal neurotransmitter release are glial cell receptors, and functional neuron–glia interactions often are mediated by these neurotransmitters. In the corpus callosum and optic nerve, for example, glutamate released by axons evokes AMPA receptor-mediated currents in NG2-positive glial precursors (3, 4), whereas axonal glutamate and ATP release leads to Ca²⁺ signals in glial cells in optic and olfactory nerves (1, 2). However, the function of these Ca²⁺ signals is still obscure. In synaptic regions, transmitter-induced Ca²⁺ signaling in astrocytes links neuronal activity to changes in blood vessel diameter (neurovascular coupling) (16), but whether neurovascular coupling occurs via glial cells other than astrocytes and in brain regions devoid of synapses, such as axon tracts, is not known. Therefore, we used olfactory ensheathing glial cells (OECs), which are closely associated with axons in the nerve layer of the olfactory bulb (17), as a monitor for glutamate and ATP to study (i) the mechanism of ectopic axonal transmitter release and (ii) the possibility of neurovascular coupling in the olfactory nerve via Ca²⁺ signaling in OECs. Our results show that ATP and glutamate release from axons is Ca²⁺-dependent and is sensitive to impairment of vesicle fusion, suggesting ectopic transmitter release via vesicles. The Ca²⁺ increase in OECs evoked by these neurotransmitters triggered vasoconstriction in the nerve layer, demonstrating that OECs are able to mediate neurovascular coupling.

Results

Release of ATP and Glutamate from Receptor Axons Stimulates Ca²⁺ Signaling in OECs.

In the superficial layers of the mouse olfactory bulb, Fluo-4 acetoxymethyl ester (Fluo-4 AM) preferentially labels olfactory ensheathing cells (OECs), which can be distinguished readily from other cells by their position in the nerve layer, their size, and their shape (2). Electrical stimulation of receptor axons in the nerve layer (individual pulse length: 1 ms; 20 Hz, 30 V) for 3 s evoked increases in the cytosolic Ca²⁺ concentration in OECs, reflected by an increase in fluorescence of 42.5 ± 3.1% ΔF (n = 49) (Fig. 1A). A single stimulation pulse evoked a Ca²⁺ rise of 30.8 ± 8.0% ΔF in 27% of the cells that responded to a 3-s train (n = 26), indicating that some OECs are able to detect the amount of neurotransmitter that is released during a single compound action potential in the nerve layer (Fig. S1). Stimulation for 3 s elicited inward currents in mitral cells of 806.7 ± 257.1 pA (n = 7); the stimulation-evoked responses in mitral cells and OECs were suppressed in the presence of tetrodotoxin (Fig. S1). After type 1 metabotropic glutamate receptors (mGluR1s) were blocked with 7-hydroxyiminocyclopropan[b]chromen-1a-carboxylic acid ethyl ester (CPCCOEt) (100 μM) or P2Y1 receptors were blocked with N(6)-methyl-2′-deoxyadenosine-3′,5′-bisphosphate (MRS2179) (60 μM) (Fig. S2), the stimulation-induced Ca²⁺ increase in OECs was reduced by 53.3 ± 4.6% (n = 49; P < 0.005) and 57.1 ± 4.1% (n = 33; P < 0.005), respectively (Fig. 1A). When both types of receptors were blocked simultaneously, the stimulation-induced Ca²⁺ transients in OECs were suppressed entirely (n = 64), indicating that glial responses are mediated by both glutamatergic and purinergic receptors. This result suggests that glutamate and ATP are coreleased by receptor axons during stimulation.

Fig. 1.

Glutamate and ATP release from olfactory receptor axons. (A) Electrical stimulation of receptor axons (20 Hz, 30 V, 3 s; indicated by an arrowhead) evokes Ca²⁺ transients in OECs that are reduced by CPCCOEt (100 μM) and are completely blocked by additional application of MRS2719 (60 μM). The response recovers after wash out of the drugs (Wash).(B) Membrane current evoked in a P2X2 receptor-expressing HEK293 cell (“sniffer cell”) by puff application (arrowhead) of ATP (100 μM). (C) Single-pulse stimulation of receptor axons evokes an inward current in a sniffer cell seeded on top of an axon bundle in the nerve layer. Arrowheads indicate stimulation artefacts. (Inset) Expanded time scale showing onset of the current response evoked by a single stimulation pulse. (D) Ionic current induced in a sniffer cell by a train of five stimulation pulses. (E) Inward currents in sniffer cells are inhibited by PPADS.

To verify the release of ATP from olfactory receptor axons, we used P2X2 receptor-expressing HEK293 cells as “sniffer cells.” Puff application of ATP (100 μM) onto sniffer cells evoked an inward current of 382 ± 63 pA (n = 6), confirming the expression of ATP-sensitive receptors (Fig. 1B). To detect ATP released from axons, sniffer cells were seeded onto exposed axon bundles of whole isolated olfactory bulbs, and whole-cell currents were recorded. Electrical stimulation (30 V) of an axon bundle attached to the sniffer cell induced inward currents (Fig. 1C), whereas stimulation of adjacent axon bundles not directly attached to the sniffer cell did not evoke currents. This observation and the short delay between stimulus and current onset (Fig. 1C, Inset) suggest fast release of ATP from receptor axons and limited diffusion of ATP. The amplitudes of stimulation-induced currents in sniffer cells increased with the number of stimulation pulses (Fig. 1 C–E). A single stimulation pulse evoked a current of 27 ± 9 pA (n = 7), whereas trains of 5 and 10 pulses (20 Hz) evoked currents of 128 ± 14 pA (n = 12) and 322 ± 55 pA (n = 9), respectively. The current amplitude evoked by 10 stimulation pulses was in the same range as the current amplitude following puff application of 100 μM ATP, suggesting that the extracellular concentration of ATP reached at least several tens of micromoles upon electrical stimulation of receptor axons. In the presence of the P2X receptor antagonist pyridoxal-phosphate-6-azophenyl-2′,4′-disulfonate (PPADS), the stimulation-evoked currents were reduced by 95.1 ± 2.4% (n = 5; P < 0.005), confirming that they were mediated by P2X2 receptors activated by ATP released upon electrical stimulation of receptor axons (Fig. 1E).

Neurotransmitter Release Is Ca²⁺ Dependent.

To test whether the release of glutamate and ATP is Ca²⁺ dependent, we suppressed voltage-gated Ca²⁺ influx into receptor axons by the l-type Ca²⁺ channel blocker diltiazem (50 μM). The efficacy of diltiazem in blocking Ca²⁺ influx was verified by measurements of Ca²⁺ changes in Calcium Green-1 dextran-filled receptor axons in the nerve layer (Fig. 2A). In the presence of diltiazem, Ca²⁺ transients induced in receptor axons by electrical stimulation were reversibly reduced by 94.8 ± 1.5% (n = 82; P < 0.005), as compared with the control stimulation in the absence of diltiazem, indicating voltage-dependent Ca²⁺ influx into axons (Fig. 2B). OECs, in contrast, appear not to express voltage-gated Ca²⁺ channels, because depolarization did not induce Ca²⁺ influx into OECs (Fig. S3A). Blocking axonal Ca²⁺ influx with diltiazem (Fig. 2C) or by withdrawal of external Ca²⁺ (Fig. S3B) reduced stimulation-induced Ca²⁺ transients in OECs by 89.3 ± 2.4% (n = 45; P < 0.005) and 96.7 ± 1.0% (n = 23; P < 0.005), respectively, as compared with control. The amplitude of ATP-evoked Ca²⁺ transients in OECs was unaffected by diltiazem. The results suggest that extrasynaptic release of ATP and glutamate along olfactory receptor axons depends on Ca²⁺ influx via l-type voltage-gated Ca²⁺ channels, in contrast to synaptic transmitter release, which depends on N- and P/Q-type Ca²⁺ channels (18).

Fig. 2.

ATP and glutamate release is Ca²⁺-dependent but hemichannel- independent. (A) Electrical stimulation (20 Hz, 30 V, 3 s) induces Ca²⁺ influx into receptor axons through diltiazem (Dilt)-sensitive Ca²⁺ channels. (B) Diltiazem almost entirely blocks stimulation-evoked Ca²⁺ transients in receptor axons. ***P < 0.005. (C) Diltiazem suppresses Ca²⁺ signaling in OECs induced by stimulation of receptor axons, whereas the amplitude of Ca²⁺ transients evoked by bath application of ATP is not affected. (D) Carbenoxolone (CBX) (100 μM) has no effect on stimulation-induced Ca²⁺ transients in OECs. (E) Ca²⁺ signaling in OECs is not affected by carbenoxolone (+Cbx). ns, not significant. In B and E, n = number of cells investigated.

Carbenoxolone blocks gap junction hemichannels and also inhibits VRAC- and P2X7 receptor-mediated neurotransmitter release (19, 20). To test the involvement of gap junction hemichannels, VRACs, and P2X7 receptors in ATP and/or glutamate release from olfactory receptor axons, we incubated the slices in carbenoxolone (100 μM). Carbenoxolone had no effect on the Ca²⁺ transients in OECs evoked by the electrical stimulation of receptor axons, suggesting that gap junctional hemichannels, VRACs, and P2X7 receptors are not involved in axonal transmitter release (Fig. 2 D and E).

Localization of Synaptic Proteins and Vesicles in Receptor Axons.

The fusion of synaptic vesicles with the plasma membrane involves numerous proteins (21). If glutamate and ATP are released from olfactory receptor axons by vesicle exocytosis, these proteins should be located not only in the glomeruli, where olfactory receptor axons terminate and synapse onto olfactory bulb neurons (Fig. 3A), but also in the nerve layer. Therefore, we stained olfactory bulb slices with antibodies against the markers of the synaptic vesicle fusion machinery, synaptophysin and bassoon, and the vesicular glutamate transporter VGluT2. Antibodies against these markers predominantly labeled olfactory glomeruli (Fig. 3 B–D). In addition, significant antibody labeling was found in the nerve layer, indicating that these proteins were located not only in the synaptic terminals of receptor axons but also along the axons, a prerequisite for vesicular release of glutamate and ATP along olfactory receptor axons. The hypothesis of vesicular neurotransmitter release from axons also is supported by the presence of vesicle-like structures in olfactory receptor axons in the direct vicinity of OECs in electron micrographs (Fig. 3 E and F). However, we did not find synaptic specializations such as active zones and postsynaptic densities in electron micrographs, suggesting that glutamate and ATP are released along olfactory receptor axons in a paracrine manner rather than at synaptiform structures.

Fig. 3.

Synaptic vesicle markers in olfactory receptor axons. (A) Olfactory receptor axons (arrowhead) traced with 4-[4-(dihexadecylamino)styryl]-N-methylpyridinium iodide (DiA) (green) in the nerve layer (NL) terminate in glomeruli (asterisks) in the glomerular layer (GL). Nuclei are labeled with propidium iodide (PI) (red). (B) Synaptophysin immunoreactivity (green) in the glomeruli (asterisk), in the glomerular layer (GL), and in axons (arrowheads) in the nerve layer (NL). (C) The vesicular glutamate transporter VGLUT2 (green) is detected in glomeruli (asterisks) and in axon bundles (arrowheads). (D) Bassoon immunoreactivity (green) is located in the glomeruli (asterisks) and in the nerve layer (arrowhead). (Scale bar: A–D, 25 μm.) (E) Electron micrograph of the nerve layer. The cell body of an OEC is highlighted in blue. (Scale bar: 0.5 μm.) (F) Region indicated by the square in E at higher magnification. Vesicle-like structures (arrowheads) are visible in olfactory receptor axons adjacent to an OEC. (Scale bar: 100 nm.)

ATP and Glutamate Are Released from Vesicles.

The presence of vesicles and vesicle-associated proteins in olfactory receptor axons cannot be considered a definite indicator of vesicular neurotransmitter release in the nerve layer, because the vesicles could be transport vesicles carrying synaptic proteins to the axon terminals. To verify the presence of functional, releasable vesicles that can undergo vesicle fusion, we used a mouse strain in which olfactory receptor axons express the vesicle fusion marker synapto-pHluorin (spH) (Fig. 4A) (22). Changes in spH fluorescence correlate linearly with the spike rate and hence with transmitter release in receptor axons (23). Electrical stimulation of olfactory receptor axon bundles (20 Hz, 30 V, 1 s) resulted in an increase in spH fluorescence of 15.0 ± 1.4% (n = 59) in the glomeruli, an increase that is comparable to spH signals evoked by intense odor stimulation (23). In axon bundles, stimulation evoked spH signals of 10.1 ± 1.2% (n = 22), indicative of vesicle fusion in axons (Fig. 4B). The spH signals in glomeruli and axon bundles were suppressed in Ca²⁺-free, EGTA-buffered saline (P < 0.005; Fig. 4C) and by 1 μM tetrodotoxin (TTX) (P < 0.005) (Fig. S1D). These results indicate Ca²⁺-dependent vesicle fusion not only in the synaptic regions in the glomeruli but also along the axons in the nerve layer.

Fig. 4.

ATP and glutamate release from receptor axons depends on vesicle fusion. (A) Expression of the vesicle fusion marker spH in receptor axon bundles (arrowheads) and glomeruli (asterisks). The tip of the stimulation pipette is depicted in yellow (arrow). (Scale bar: 50 μm.) (B) Pseudocolor overlay indicating regions of spH fluorescence increase upon electrical stimulation of axons. Warm colors represent large fluorescence increases; cool colors represent small fluorescence increases. (C) Electrical stimulation (Stim, arrowhead) (20 Hz, 30 V, 1 s) of receptor axons results in Ca²⁺-dependent increases in spH fluorescence in glomeruli and axons. (D) Ca²⁺ transients evoked by axon stimulation (20 Hz, 30 V, 3 s) were greatly reduced after incubation of OECs with bafilomycin A1 (10 μM) for 30 min or (E) with botulinum toxin A (150 μg/mL) for 40 min.

Vesicular release of neurotransmitters can be suppressed by preventing the loading of vesicles with transmitter molecules using bafilomycin A1 or by cleaving the soluble N-ethylmaleimide sensitive factor attachment protein receptor (SNARE) complex required for vesicle fusion using botulinum toxin A (24, 25). We monitored transmitter release evoked by receptor axon stimulation via Ca²⁺ changes in OECs before and after incubation of olfactory bulb slices with either 10 μM bafilomycin A1 for 30 min or 150 μg/mL botulinum toxin A for 40 min (Fig. 4 D and E). Stimulation-induced Ca²⁺ transients were reduced after incubation with bafilomycin A1 and botulinum toxin by 87.2 ± 2.9% (n = 63; P < 0.005) and 92.6 ± 1.1% (n = 82; P < 0.005), respectively, whereas ATP- and dihydroxyphenylglycin (DHPG)-induced Ca²⁺ signaling was not affected or was affected only weakly (Fig. S4). These results demonstrate that release of both ATP and glutamate from olfactory receptor axons depends on functional vesicles and vesicle fusion.

Glial Ca²⁺ Signaling Mediates Vasoconstriction.

Cerebral blood flow in gray matter has been shown to be controlled by Ca²⁺ signaling in astrocytes excited by synaptic release of neurotransmitters (16). We were interested in whether glial cells in the olfactory nerve, where neurotransmitters are released in the absence of synaptic structures, also can mediate neurovascular coupling. Fig. 5 A and B demonstrates the tight morphological relationship between blood vessels and OECs, which ensheath blood vessels with their processes. We bulk-loaded OECs in whole isolated olfactory bulbs with Fluo-4 AM using multicell bolus loading (26) and visualized blood vessels by injecting sulforhodamine 101 (SR101) (Fig. S5). In OECs, puff application of ATP (1 mM) evoked Ca²⁺ transients with a mean amplitude of 203.6 ± 10.7% ΔF (n = 64), which were followed by a constriction of adjacent blood vessels by 33.9 ± 4.7% (n = 30) (Fig. 5 C and D and Movie S1). When olfactory bulbs were preincubated in MRS2179 (100 μM), ATP-induced Ca²⁺ transients in OECs were reduced by 79% (n = 32; P < 0.005), and failed to evoke vasoconstriction (n = 13) (Fig. 5E). After washout of MRS2179, ATP evoked Ca²⁺ signaling and vasoconstriction recovered.

Fig. 5.

Ca²⁺ signaling in OECs mediates vasoconstriction. (A) Immunostaining against S100 calcium-binding protein (OECs, green) and mouse IgG [blood vessels (red)]. Nuclei are counterstained with Hoechst (blue). The arrowhead indicates a blood vessel that is depicted at higher magnification in B. GL, glomerular layer; NL, nerve layer. (Scale bar: 50 μm.) (B) OEC somata (asterisks) are located in the direct vicinity of blood vessels, and OEC processes ensheath blood vessels (arrowheads). (Scale bar: 10 μm.) (C) Time series of Fluo-4–loaded OECs (green) and SR101-injected blood capillaries (red). ATP (1 mM) was applied to OECs via a puff pipette. Fluo-4 was measured in region of interest 1 (ROI 1), and SR101 was measured in ROI 2 (Movie S1). (Scale bar: 15 μm.) (D) Time course of a Ca²⁺ transient (upper trace) and the SR101 fluorescence (lower trace, to visualize vasoconstriction). (E) Analysis of Ca²⁺ response in OECs (Fluo-4) and vasoconstriction evoked by ATP in ACSF (ATP), in the presence of 100 μM MRS2179 (ATP+MRS) and after washout of MRS2179 (ATP wash). MRS2179 reduced ATP-mediated Ca²⁺ transients in OECs and completely inhibited vasoconstriction. ***P > 0.005. (F) Fluorescence of Fluo-4 (green) and SR101 (red) before (Left) and after (Right) laser photolysis of NP-EGTA in OECs by illumination of the entire field of view with a 405-nm laser (Movie S2). (G) Time course of Ca²⁺ transients in OECs (upper trace) and vasoconstriction (lower traces) upon photolysis of NP-EGTA (arrow). (H) Spot illumination with 405 nm results in a Ca²⁺ transient in a single OEC (Movie S3). (I) Time course of Ca²⁺ transients in OECs (upper trace) and vasoconstriction (lower traces) upon photolysis of NP-EGTA (arrow). (J) Ca²⁺ response in OECs evoked by electrical stimulation (20 Hz, 30 V, 3 s) of axons (Movie S4). (K) Time course of Ca²⁺ transients and vasoconstriction upon axonal stimulation. (Scale bars in F, H, and J: 10 μm.) The rapid drop in SR101 fluorescence (arrow) is caused by a nonfluorescent erythrocyte that became stuck in the ROI in the blood vessel.

To test whether a Ca²⁺ increase in OECs is sufficient to trigger vasoconstriction, we released “caged Ca²⁺” in NP-EGTA–loaded OECs by photolysis. When the entire field of view was illuminated with 405-nm light, Ca²⁺ transients of 181.1 ± 22.5% ΔF (n = 23) were evoked in several OECs (Fig. 5F), leading to a global constriction of blood vessels of 19.0 ± 2.6% (n = 16) (Fig. 5G and Movie S2). When the 405-nm illumination was focused on a 2 × 2 μm spot, Ca²⁺ was elevated only in single OECs (amplitude 149.0 ± 30.6% ΔF; n = 7), followed by a local constriction of adjacent blood vessels by 20.0 ± 2.8% (n = 10; Fig. 5 H and I and Movie S3). When the illumination spot was located outside the OEC, or when OECs not loaded with NP-EGTA were recorded, 405-nm illumination failed to evoke Ca²⁺ transients in OECs, and no vascular response could be measured, ruling out nonspecific vascular effects by 405-nm illumination per se or out-of-focus photolysis of NP-EGTA (e.g., in axons or epithelial cells) (Fig. S6). In addition, Ca²⁺ transients in OECs evoked by axon stimulation (20 Hz, 30 V, for 3 s) resulted in vasoconstriction of 38.2 ± 4.1% (n = 10; P < 0.005) (Fig. 5 J and K and Movie S4), suggesting that axonal activity can elicit neurovascular coupling in the olfactory nerve layer, mediated by axon–glia communication and glial Ca²⁺ signaling.

Discussion

Two recent studies showed that axons in white matter not only conduct action potentials but also release the neurotransmitter glutamate via vesicles at synaptiform structures and hence participate in information processing (3, 4). Both studies examined axons in the corpus callosum, where they demonstrate vesicular release of glutamate. It remained unclear, however, whether vesicular release is a general phenomenon in white matter throughout the brain or is restricted to the corpus callosum as a unique conductive structure transmitting a huge amount of information between the cortical hemispheres (27). In addition, it is not known whether neurotransmitters other than glutamate can be released ectopically from vesicles along axons. In axons in the nerve layer of the olfactory bulb investigated in the present study, we demonstrated vesicular release not only of glutamate but also of ATP, which is an important cotransmitter (28, 29). In electron micrographs we did not find synaptic specializations such as postsynaptic densities, and immunoreactivity of proteins involved in vesicle fusion was distributed evenly within the olfactory nerve layer, suggesting that olfactory receptor axons do not contain active zones. This even distribution is in contrast to the axons in the corpus callosum, where punctate immunoreactivity against the vesicular glutamate transporter VGluT1 was demonstrated and where junctions between axons and NG2-positive cells possess synaptic specializations (4).

ATP is an important mediator in neuron–glia communication (30, 31). ATP release from extrasynaptic sites involved in neuron–glia interactions has been reported. For example, in dorsal root ganglion cells, Zhang et al. (32) described a quantal and Ca²⁺-dependent release of ATP from somata of acutely dissociated dorsal root ganglion cells; this ATP subsequently stimulated adjacent satellite glial cells. Stevens and Fields (33) and Stevens et al (34) reported release of ATP from axons of cultured dorsal root ganglion cells, causing Ca²⁺ signals in cocultured Schwann cells and oligodendrocyte precursors and thus inhibiting proliferation and maturation of Schwann cells but promoting myelination by oligodendrocyte precursors. This result emphasizes the significance of axonal ATP release for the control of myelination (30). Glial cells themselves also can release ATP to mediate axon–glia communication. In rat optic nerve, mechanical and glutamatergic stimulation of astrocytes leads to the release of ATP from astrocytes, triggering Ca²⁺ waves covering adjacent astrocytes and oligodendrocytes (35). In the present study, glutamate-dependent ATP release from OECs upon stimulation of receptor axons appears not to be the primary source of ATP, because the ATP-mediated Ca²⁺ response in OECs remained when glutamate receptors were inhibited (Fig. 1A). Only if both mGluRs and P2Y1 receptors were blocked was the stimulation-evoked response entirely suppressed, indicating that the glutamatergic and the purinergic pathways proceed in parallel and not successively. Thus, simultaneous release of glutamate and ATP from receptor axons is proposed here, as has been shown at olfactory receptor axon terminals in the neuropil of the olfactory bulb (36). The prominent role of the purinergic system in the olfactory bulb, and in particular in the nerve layer, is emphasized by the high expression of ATP-degrading enzymes in the olfactory nerve layer (37).

Neuronal activity has been shown to be linked to changes in blood flow, a mechanism termed “neurovascular coupling” (16). The mechanism of neurovascular coupling in the olfactory bulb is controversial, because changes in blood flow of periglomerular blood capillaries has been shown to be triggered in an astrocyte-dependent as well as astrocyte-independent manner (23, 38). In addition, differences in astrocyte-mediated neurovascular coupling have been found between in vivo and in vitro preparations. Activation of Ca²⁺ transients in astrocytes of the somatosensory cortex and the olfactory bulb in vivo by stimulation of afferents was accompanied by vasodilation and an increase in blood flow (23, 39, 40). In brain-slice preparations, the situation appears more complicated, and both vasodilation and vasoconstriction could be measured (39, 41, 42), depending on the metabolic state of the tissue and the amplitude of astrocytic Ca²⁺ transients (43, 44). In all studies mentioned above, glial cells are located in the neuropil and couple neuronal activity to responses of the vasculature via detection of synaptically released neurotransmitters. It was not known, however, whether glial Ca²⁺ signaling evoked by ectopic release of neurotransmitters is used also to regulate blood flow in the brain. The vasoconstriction mediated by Ca²⁺ transients in OECs, which can be evoked by ectopic release of ATP and glutamate from receptor axons, demonstrates glial control of blood vessels in an axon tract devoid of synapses. Hence, neurovascular coupling probably is a major function of glial Ca²⁺ signaling not only in gray matter but also in white matter regions in the brain.

Methods

Preparation of Brain Slices and Whole Olfactory Bulbs.

Olfactory bulbs were obtained from NMRI mice from postnatal days 0 to 17 (P0–P17) as described previously (2). Animals were decapitated, and the olfactory bulbs were quickly removed from the brain. For whole-bulb preparations, olfactory bulbs were glued to round coverslips, placed in Ca²⁺-reduced artificial cerebrospinal fluid (ACSF), and allowed to recover at 30 °C for 1 h. Olfactory bulbs were used within 3 h after preparation. To check the viability of OECs in whole olfactory bulbs, which might be impaired after storage for several hours, we bath-applied ATP at different times after preparation and did not find any sign of impairment in terms of generation and back-regulation of ATP-induced Ca²⁺ rises after storage up to 6 h. For brain slices, olfactory bulbs were glued to the stage of a Leica VT 1000s vibroslicer. Sagittal slices 250 μm thick were cut in cooled, Ca²⁺-reduced ACSF and were stored at 30 °C for 1 h before the experiments.

Calcium Imaging.

For bulk dye loading in brain slices, olfactory bulb slices were incubated in ACSF containing 2 μM Fluo-4 AM for 1 h at room temperature. Dye-loaded slices were transferred into fresh ACSF and kept at room temperature for up to 6 h. In the nerve layer, Fluo-4 AM preferentially loads OECs (2). OECs in whole olfactory bulbs were stained by multicell bolus loading (26). ACSF containing 200 μM Fluo-4 AM was placed in a micropipette that was inserted into the nerve layer under visual control. Fluo-4 AM was ejected by pressure of 0.4 bar for 10 s. After 30 min OECs were brightly labeled by Fluo-4. In some experiments, 1 mM NP-EGTA AM (caged Ca²⁺) was added to the Fluo-4 AM solution to coload OECs with Fluo-4 and NP-EGTA. Uncaging of NP-EGTA was achieved by illumination with a 405-nm laser diode at 5 mW for 1–2 s.

Slices and whole olfactory bulbs were fixed in a perfusion chamber and continuously superfused with ACSF. Ca²⁺ imaging was performed using confocal laser scanning microscopy (Zeiss LSM 510 and Nikon EC1). Changes in the intracellular calcium concentration of OECs were induced either by bath or puff application of ATP and DHPG or by electrical stimulation of olfactory receptor axons using a stimulation pipette (tip resistance ∼0.5–2 MΩ).

Data Analysis.

Changes in intracellular Ca²⁺ as well as changes in spH fluorescence are given as changes in fluorescence (ΔF) relative to the basal fluorescence at the beginning of an experiment, which was normalized to 100%. All values are given as means ± SEM. The number of evaluated cells is indicated by n. For each experiment and protocol, data from at least three animals were analyzed. Significance of statistical difference was calculated using Student's t test for large data sets (n > 30) and the Mann–Whitney U test for small data sets at 95% confidence level (P ≤ 0.05). Exact P values are given between 0.05 and 0.005; P < 0.005 is stated for P values smaller than 0.005.

Further information is given in SI Methods.