Ca²⁺-permeable AMPA receptors in mouse olfactory bulb astrocytes

Abstract

Ca²⁺ signaling in astrocytes is considered to be mainly mediated by metabotropic receptors linked to intracellular Ca²⁺ release. However, recent studies demonstrate a significant contribution of Ca²⁺ influx to spontaneous and evoked Ca²⁺ signaling in astrocytes, suggesting that Ca²⁺ influx might account for astrocytic Ca²⁺ signaling to a greater extent than previously thought. Here, we investigated AMPA-evoked Ca²⁺ influx into olfactory bulb astrocytes in mouse brain slices using Fluo-4 and GCaMP6s, respectively. Bath application of AMPA evoked Ca²⁺ transients in periglomerular astrocytes that persisted after neuronal transmitter release was inhibited by tetrodotoxin and bafilomycin A1. Withdrawal of external Ca²⁺ suppressed AMPA-evoked Ca²⁺ transients, whereas depletion of Ca²⁺ stores had no effect. Both Ca²⁺ transients and inward currents induced by AMPA receptor activation were partly reduced by Naspm, a blocker of Ca²⁺-permeable AMPA receptors lacking the GluA2 subunit. Antibody staining revealed a strong expression of GluA1 and GluA4 and a weak expression of GluA2 in periglomerular astrocytes. Our results indicate that Naspm-sensitive, Ca²⁺-permeable AMPA receptors contribute to Ca²⁺ signaling in periglomerular astrocytes in the olfactory bulb.

Introduction

It has become increasingly evident during the past decade that astrocytes are far more than supportive cells in the brain, but rather take active part in information processing such as synaptic transmission and synaptic plasticity^1,2. Most of the functions attributed to astrocytes are governed by cytosolic Ca²⁺ signaling^3,4. Astrocytes are equipped with a plethora of receptors for neurotransmitters, neuropeptides and growth factors, most of which are linked to Ca²⁺ release from internal stores⁵. Hence, it is generally accepted that internal Ca²⁺ release is the main player in glial cell physiology. Recent studies, however, challenge this notion and demonstrate pivotal roles of Ca²⁺ influx in spontaneous Ca²⁺ signaling and in astrocyte function such as neurovascular coupling^6,7. Astrocytes express Ca²⁺-permeable ion channels such as ionotropic neurotransmitter receptors, transient receptor potential channels and store-operated Ca²⁺ channels that mediate Ca²⁺ influx from the extracellular space^3,5,6,7. For example, Bergmann glial cells, specialized astrocytes in the cerebellum, possess ionotropic glutamate receptors of the AMPA type that consist of GluA1 and GluA4 subunits, but lack the GluA2 subunit and hence exhibit high Ca²⁺ permeability^8,9. Artificial insertion of the GluA2 subunit renders AMPA receptors in Bergmann glial cells Ca²⁺-impermeable and results in retraction of glial processes from synapses of Purkinje cells and abnormal synaptic currents¹⁰. This phenotype was mimicked in glia-specific GluA1/GluA4 double knock-out mice, which additionally showed impairment in fine motor coordination, emphasizing a pivotal role of Ca²⁺-permeable AMPA receptors in neuron-glia interactions¹¹. In other brain regions such as the thalamus, astrocytes possess AMPA receptors of different subunit composition including variable contribution of the GluA2 subunit to the channel assembly and hence display intermediate Ca²⁺ permeability¹². In the hippocampus, NG2 glial cells express AMPA receptors, while astrocytes lack these receptors^13,14,15,16.

In the olfactory bulb, activation of neurotransmitter receptors in astrocytes has been shown to result in Ca²⁺ transients¹⁷. The olfactory bulb is the first relay station of odor information processing and is targeted by axons of sensory neurons in the olfactory epithelium, the olfactory receptor neurons. Olfactory receptor neurons release glutamate and ATP in neuropilar regions called glomeruli^18,19, where they stimulate Ca²⁺ signaling in periglomerular astrocytes by mGluR₅ and P2Y₁ receptors^20,21. In addition, ATP is degraded to adenosine, acting on astrocytic A_2A receptors²¹. Ca²⁺ increases in olfactory bulb astrocytes have been reported to release ATP from astrocytes and to trigger vasoresponses in blood vessels contacted by astrocytic end feet^20,22,23,24. All Ca²⁺ responses to neurotransmitters measured in olfactory bulb astrocytes were mediated by Ca²⁺ release from internal stores, while Ca²⁺ influx from the extracellular space has not been demonstrated so far¹⁷. In the present study, we were interested whether olfactory bulb astrocytes express Ca²⁺-permeable AMPA receptors and whether these receptors are activated by glutamate released from olfactory receptor neurons. We studied Ca²⁺ responses and membrane currents in olfactory bulb astrocytes to application of AMPA and kainate, using Ca²⁺ imaging in brain slices and whole-cell patch-clamp recordings in acutely isolated astrocytes. Kainate-evoked membrane currents as well as Ca²⁺ transients induced by AMPA and electrical stimulation of olfactory receptor axons were partly reduced by Naspm (N-[3-[[4-[(3-aminopropyl) amino] butyl] amino] propyl]-1-naphthaleneacetamide trihydrochloride), an AMPA receptor blocker selective for GluA2-lacking, Ca²⁺-permeable AMPA receptors. Immunohistological staining revealed expression of GluA1, GluA2 and GluA4 in olfactory bulb astrocytes. The results indicate that olfactory bulb astrocytes possess both GluA2-containing and GluA2-lacking AMPA receptors, the latter being blocked by Naspm.

Results

Olfactory bulb astrocytes respond to AMPA application

We were interested whether olfactory bulb astrocytes respond to bath application of AMPA with Ca²⁺ signaling. We used application of ATP to test the viability of the cells and to identify astrocytes in the glomerular layer. Olfactory bulb astrocytes express P2Y₁ receptors and respond to bath application of ATP and ADP with Ca²⁺ transients, whereas olfactory bulb neurons do not²¹. In the present study, 100 μM ATP evoked a Ca²⁺-dependent increase in Fluo-4 fluorescence by 1.34 ± 0.08 ΔF/F₀ (n = 53) in periglomerular astrocytes, while in neurons, ATP did not evoke Ca²⁺ signaling (Fig. 1a). 59.3% of ATP-sensitive astrocytes and all neurons also responded to bath application of 50 μM AMPA (Fig. 1a). The mean Ca²⁺ increase was 1.18 ± 0.09 ΔF/F₀ (n = 53) in AMPA-responding astrocytes and 1.62 ± 0.14 ΔF/F₀ (n = 59) in neurons. To test whether AMPA evoked Ca²⁺ transients in periglomerular astrocytes directly or via activation of neurons and subsequent neurotransmitter release, we suppressed vesicular neurotransmitter release by incubation of the brain slices with 10 μM bafilomycin A1 (duration 40 min) and added 1 μM tetrodotoxin (TTX) to the artificial cerebrospinal fluid (ACSF) (Fig. 1b and c). We used NMDA as a control of successful TTX/bafilomycin treatment, since direct NMDA-induced stimulation of astrocytes is negligible and hence NMDA-evoked Ca²⁺ transients in astrocytes are expected to be suppressed by TTX/bafilomycin²². 100 μM NMDA evoked an increase in Ca²⁺ of 0.97 ± 0.07 ΔF/F₀ (n = 53) under control conditions, while the increase was significantly reduced to 0.32 ± 0.03 ΔF/F₀ (n = 82; p < 0.001) by treatment with TTX/bafilomycin (Fig. 1d). In contrast to NMDA, both ATP and AMPA induced Ca²⁺ transients in astrocytes in the presence of TTX/bafilomycin that did not differ in amplitude compared to transients evoked in the absence of TTX/bafilomycin (Fig. 1b and c). The amplitudes of the Ca²⁺ transients in the presence of TTX/bafilomycin were 1.26 ± 0.07 ΔF/F₀ (n = 82; ATP; p = 0.479) and 1.02 ± 0.13 ΔF/F₀ (n = 82; AMPA; p = 0.318). The results indicate that ATP and AMPA directly induced Ca²⁺ transients in astrocytes. Therefore, we used bafilomycin A1 and TTX in all following Ca²⁺ imaging experiments (except for electrical stimulation of axons) to isolate the direct response in astrocytes.

Figure 1: Bafilomycin A1 and TTX fail to reduce AMPA-induced Ca²⁺ transients in periglomerular astrocytes.

(a) Ca²⁺ imaging traces of a single periglomerular astrocyte (upper trace) and neuron (lower trace) showing Ca²⁺ responses to ATP (100 μM) and AMPA (50 μM). (b) Ca²⁺ responses evoked by ATP (100 μM), NMDA (100 μM) and AMPA (50 μM) in the absence and (c) presence of TTX (1 μM) and bafilomycin A1 (10 μM). (d) Normalized and averaged amplitudes of Ca²⁺ transients evoked by ATP, NMDA and AMPA under control conditions (open bars) and in the presence of bafilomycin A1 and TTX (gray bars). ***p < 0.001.

Ca²⁺-permeable AMPA receptors mediate Ca²⁺ influx into periglomerular astrocytes

AMPA can activate AMPA receptors as well as kainate receptors²⁵, both of which may trigger Ca²⁺ influx through the receptor channel itself. In addition, AMPA/kainate receptors have been shown to stimulate G protein-coupled pathways, including Ca²⁺ release from intracellular stores^26,27,28. To test whether Ca²⁺ is released from internal stores, we applied AMPA (25 μM) after Ca²⁺ stores were depleted by incubation with 20 μM cyclopiazonic acid (CPA). As shown in Fig. 2, the AMPA-induced Ca²⁺ response in periglomerular astrocytes was only slightly reduced after Ca²⁺ store depletion. In Ca²⁺-free, EGTA-buffered ACSF, however, the Ca²⁺ response was entirely blocked (n = 53; p < 0.001), indicating that AMPA induced Ca²⁺ influx from the extracellular space.

Figure 2: AMPA-induced changes in intracellular Ca²⁺ are not dependent on intracellular Ca²⁺ stores.

(a) Ca²⁺ transient of a periglomerular astrocyte evoked by AMPA (25 μM) under control conditions, in the presence of CPA (20 μM), and in Ca²⁺-free saline in CPA. (b) CPA had only a small effect on AMPA-induced Ca²⁺ transients, while Ca²⁺ transients were entirely suppressed in Ca²⁺-free saline. **p < 0.01; ***p < 0.001.

The AMPA-induced Ca²⁺ influx could be mediated by Ca²⁺-permeable AMPA receptors or by AMPA-evoked depolarization and subsequent activation of voltage-gated Ca²⁺ channels. To test whether AMPA evoked Ca²⁺ influx through Ca²⁺-permeable AMPA receptors we employed Naspm, a highly selective antagonist of GluA2-lacking, Ca²⁺-permeable AMPA receptors^29,30. In the presence of 50 μM Naspm, AMPA-evoked Ca²⁺ transients were significantly reduced by 68.0 ± 2.2% (n = 72; p < 0.001) (Fig. 3a and b). In addition, depolarization of the cells by application of 50 mM K⁺ in the presence of bafilomycin A1 and TTX elicited only a small Ca²⁺ transient in astrocytes of 0.13 ± 0.09 ΔF/F₀ (n = 16), leading to the conclusion that Ca²⁺ influx through voltage-gated Ca²⁺ channels is negligible (Fig. 3c). Hence, AMPA-evoked Ca²⁺ transients appear to be mainly mediated by Ca²⁺ influx through Ca²⁺-permeable AMPA receptors. We also used acutely isolated astrocytes to test the effect of Naspm on AMPA-evoked Ca²⁺ signaling. AMPA-evoked Ca²⁺ transients were significantly reduced by 50 μM Naspm (n = 31; p < 0.001), confirming the involvement of Ca²⁺-permeable AMPA receptors in Ca²⁺ signaling in olfactory bulb astrocytes (Fig. 3d and e). On average, 25 μM AMPA induced Ca²⁺ transients with an amplitude of 2.50 ± 0.50 ΔF/F₀ (n = 31), which were reduced to 52.7 ± 5.3% by Naspm.

Figure 3: Naspm reduces AMPA receptor-induced Ca²⁺ transients and membrane currents in periglomerular astrocytes.

(a) Effect of Naspm (50 μM) on AMPA-induced Ca²⁺ transients after incubation of brain slices in bafilomycin A1 and TTX. (b) Naspm significantly reduced AMPA-induced Ca²⁺ transients in periglomerular astrocytes. Wash out of Naspm led to a significant recovery of the Ca²⁺ response. (c) Increasing external K⁺ to 50 mM evoked large Ca²⁺ transients in neurons, but only small Ca²⁺ rises in astrocytes, indicating lack of voltage-gated Ca²⁺ influx in astrocytes. (d) Fluo-4-loaded isolated astrocyte. Scale bar: 20 μm. (e) Effect of Naspm (50 μM) on AMPA-evoked Ca²⁺ transients in an isolated olfactory bulb astrocyte. (f) Immunostaining of an eGFP-positive periglomerular astrocyte in an hGFAP-eGFP mouse (anti-GFP, green) and colabeling of GFAP as a marker for astrocytes (anti-GFAP, red). Scale bar: 10 μm. (g) Whole-cell current trace of a dissociated astrocyte recorded in BaCl₂ (100 μM), quinine (100 μM) and cyclothiazide (100 μM). Kainate (500 μM) evoked an inward current that was entirely blocked by GYKI 53655 (100 μM), but (h) was only partly reduced by Naspm (50 μM). (i) Normalized averaged effects of GYKI 53655and Naspm on kainate-induced currents.

We aimed to verify these results in electrophysiological experiments on acutely dissociated astrocytes from the olfactory bulb glomerular layer, using hGFAP-eGFP reporter mice to identify astrocytes (Fig. 3f). We applied 500 μM kainate in the presence of 100 μM cyclothiazide to induce AMPA receptor-mediated membrane currents with minimal desensitization^12,31. We and others have previously shown that activation of AMPA/KA receptors in glial cells leads to Na⁺ influx that plugs K⁺ channels^32,33. To avoid that the kainate-induced receptor currents were obscured by a simultaneous block of Kir channels, BaCl₂ was applied. Kainate application evoked an inward current of 427.8 ± 190.3 pA (n = 15) (at −70 mV) corresponding to a current density of 16.0 ± 5.1 pA/pF (n = 15) (Fig. 3f). 100 μM GYKI 53655 (1-(4-aminophenyl)-3-methylcarbamyl-4-methyl-3,4-dihydro-7,8-methylenedioxy-5H-2,3-benzodiazepine hydrochloride), a selective AMPA receptor blocker barely acting on kainate receptors³⁴, entirely inhibited the kainate-induced inward current (n = 11; p < 0.001), confirming the activation of AMPA receptors by kainate in our experiments (Fig. 3g and i). In contrast to GYKI 53655, Naspm (50 μM) had variable effects on the kainate-induced inward current. In 2 out of 8 experiments, Naspm did not affect the kainate-evoked inward current, while in one experiment, Naspm entirely blocked the kainate-induced inward current. In the remaining 5 experiments, Naspm reduced the current by different amounts. In Naspm-sensitive cells, the kainate-induced inward current was significantly reduced by Naspm to 37.4 ± 13.7% of the control (n = 6; p = 0.031) (Fig. 3h and i). These results show that AMPA induced an influx of Ca²⁺ into periglomerular astrocytes that in the majority (75%) of cells is Naspm-sensitive and thus mediated by Ca²⁺-permeable AMPA receptors.

Distribution of AMPA receptor subunits in the glomerular layer

The Ca²⁺ permeability of AMPA receptors critically depends on the relative abundance of GluA2 within the channel complex^14,35,36. Since our electrophysiological and Ca²⁺ imaging data suggest Ca²⁺-permeable AMPA receptors in periglomerular astrocytes, we investigated the cellular distribution of GluA2 in the olfactory bulb. GluA2 immunoreactivity was mainly found rather homogeneously in the external plexiform layer and in somata of periglomerular cells, whereas GluA2 immunoreactivity was much weaker in the synaptic neuropil in the core of the glomeruli (Fig. 4a). Only moderate colocalization with GFAP-positive periglomerular astrocytes was detected (Fig. 4a, merged image). We also investigated the distribution of GluA1 and GluA4. GluA3 was not investigated because it is barely expressed in the olfactory bulb^37,38. GluA1 was widely distributed in the glomerular and the external plexiform layers, including the neuropil of glomeruli (Fig. 4b). Intense colocalization of GluA1 and GFAP was found in periglomerular astrocytes and their processes in the neuropil. GluA4 immunoreactivity resembled GluA1 immunoreactivity, with universal distribution in the external plexiform layer, the glomerular layer and the glomerular neuropil as well as clear colocalization with GFAP (Fig. 4c). We used cerebellar tissue of the same animals to verify specificity of antibody staining. The distribution of all three subunits investigated was in accordance with the distribution found in previous publications^39,40,41,42; GluA1 was expressed in both neurons and glial cells, GluA2 exclusively in neurons and GluA4 predominantly in Bergmann glial cells (Supplementary Fig. S1).

Figure 4: Immunostaining of AMPA receptor subunits in the olfactory bulb.

(a) GluA2 immunoreactivity (green) was detected in the external plexiform layer (EPL) and in cell bodies surrounding glomeruli. Glomeruli are indicated by asterisks. Moderate GluA2 immunoreactivity was also found in astrocytes highlighted by GFAP immunoreactivity (red), as indicated by yellow pixels in the merged image. Arrows point to astrocyte structures that were colabeled with GluA immunoreactivity. Nuclei were stained with Hoechst 33342 (blue). (b) GluA1 and GFAP colocalization. (c) GluA4 and GFAP colocalization. Scale bars: 20 μm.

Endogenous glutamate release activates astrocytic Ca²⁺-permeable AMPA receptors

Olfactory receptor neurons (ORN) release glutamate from their axon terminals in the glomeruli, which evokes Ca²⁺ signaling in astrocytes^18,20,23. We studied the effect of glutamate released upon electrical stimulation (10 Hz, 2 s) of ORN axons on astrocytic Ca²⁺ (Fig. 5). We crossbred Glast-CreERT2 and GCaMP6s^fl/fl mice to receive mice in which the genetic Ca²⁺ indicator GCaMP6s is specifically expressed by astrocytes. In brain slices of these mice, ORN stimulation resulted in Ca²⁺ transients with a mean amplitude of 3.51 ± 0.25 ΔF/F₀ (n = 51) in periglomerular astrocytes. We aimed to isolate the putative Ca²⁺ response induced by activation of Ca²⁺-permeable AMPA receptors in astrocytes by applying a mix of receptor blockers antagonizing metabotropic neurotransmitter receptors known to induce Ca²⁺ transients in olfactory bulb astrocytes. In addition, we reduced glutamatergic activation of neurons and hence indirect effects with D-APV (D-2-amino-5-phosphonovaleric acid). This blocker mix significantly reduced the stimulation-induced Ca²⁺ response in periglomerular astrocytes to 56.4 ± 3.0% of the control (n = 51; p < 0.001) (Fig. 5a). Addition of Naspm (50 μM) further decreased the amplitude of stimulation-induced Ca²⁺ transients to 24.4 ± 2.2% of the control (n = 51). Ca²⁺ transients in the presence of Naspm were significantly smaller as compared to Ca²⁺ transients in the presence of the mix of blockers without Naspm (p < 0.001), indicating that glutamate release from OSN axons triggers Ca²⁺ signaling in periglomerular astrocytes via Ca²⁺-permeable AMPA receptors.

Figure 5: Electrical stimulation of axons of olfactory receptor neurons evokes Naspm-sensitive Ca²⁺ transients in GCaMP6s-expressing astrocytes.

(a) Ca²⁺ transients evoked by electrical stimulation in the absence of receptor blockers, in the presence of a mix of receptor blockers (2 μM MPEP, antagonist of mGluR₅; 30 μM MRS2179, P2Y₁ receptor antagonist; 100 μM D-APV, NMDA receptor antagonist; 1 μM SCH442416, type 1 dopamine receptor antagonist; 100 μM CPCCOET, mGluR₁ antagonist; 2 μM SCH23390, A_2A receptor antagonist) and after addition of Naspm (50 μM). Inset: First Ca²⁺ response (control) at larger time scale. Inset scale bars: 20 s, 200% ΔF/F₀. (b) Normalized averaged amplitudes of Ca²⁺ rises evoked by electrical stimulation.

Discussion

In the present study, we have investigated the role of AMPA receptors in Ca²⁺ signaling in periglomerular astrocytes of the olfactory bulb. Our results show that axonal stimulation activates AMPA receptor-mediated Ca²⁺ influx into periglomerular astrocytes. This Ca²⁺ influx was largely reduced by Naspm, a blocker of GluA2-lacking, Ca²⁺-permeable AMPA receptors^29,30. Our antibody staining revealed expression of GluA1, GluA2 and GluA4 in periglomerular astrocytes, and patch-clamp recordings demonstrated only partial block of AMPA receptor-mediated inward currents by Naspm, suggesting that both GluA2-containing and GluA2-lacking AMPA receptors are expressed by periglomerular astrocytes.

Abundant expression of GluA1, GluA2 and GluA4 in the rodent olfactory bulb has been published before. GluA1, e.g., is mainly found in the external plexiform and the glomerular layers and is expressed by periglomerular neurons and mitral/tufted cells^39,43,44,45. GluA2 has the highest expression of all GluA subunits in the olfactory bulb as assed by PCR³⁶ and is located in mitral/tufted cells and granule cells⁴⁰. GluA3 expression in the olfactory bulb is negligible, while GluA4 expression is moderate and GluA4 protein could be detected in the mitral cell layer, the external plexiform layer, the glomerular layer and the nerve layer^38,40. However, all these studies focused on neurons, and none of them investigated the colocalization of the GluA subunits with astrocytes. We colabeled GluA immunostainings with an antibody against GFAP to highlight astrocytes in the glomerular layer. Colocalization of GluA subunits and GFAP was found for all subunits tested, with obvious colocalization for GluA1 and GluA4, but only moderate colocalization for GluA2. This suggests that only a fraction of AMPA receptors in periglomerular astrocytes comprise GluA2 subunits, while the remaining fraction of AMPA receptors lacks GluA2 subunits and hence is both Ca²⁺-permeable and Naspm-sensitive^35,36,46. This is confirmed by the partial block of AMPA receptor currents by Naspm as measured in acutely dissociated, GFP-labeled astrocytes.

Several results indicate that in periglomerular astrocytes, AMPA directly gates Ca²⁺ influx. Firstly, AMPA-induced Ca²⁺ transients upon agonist bath application were not reduced by suppressing potential indirect effects (through neuronal transmitter release) with the sodium channel blocker TTX and bafilomycin, an inhibitor of vesicular H⁺ pumps required for filling synaptic vesicles with neurotransmitter molecules⁴⁷. NMDA-evoked and high-K⁺-evoked Ca²⁺ transients in astrocytes, in contrast, were greatly reduced by TTX/bafilomycin, in line with the notion that olfactory astrocytes do not significantly express voltage-gated Ca²⁺ channels and NMDA receptors and hence high-K⁺-evoked and NMDA-evoked Ca²⁺ signaling was mainly due to neuronal transmitter release. Whether the remaining, TTX/bafilomycin-insensitive part of NMDA-evoked Ca²⁺ transients reflects insufficient efficacy of TTX/bafilomycin or expression of NMDA receptors in olfactory bulb astrocytes, as demonstrated for cortical astrocytes and for oligodendrocytes, remains to be shown^{48,49,50,51,52,53}. In addition, TTX/bafilomycin-insensitive neurotransmitter release such as reversal of neurotransmitter uptake due to the NMDA-evoked increase in Na⁺ in neurons and subsequent activation of astrocytic receptors might also contribute to NMDA-evoked Ca²⁺-transients in astrocytes. Such TTX/bafilomycin-insensitive neurotransmitter release could contribute to astrocytic Ca²⁺ signalling not only upon application of NMDA, but also AMPA. However, it is very unlikely that this has a major contribution to the AMPA-evoked Ca²⁺ signaling, since the same mechanism is activated during NMDA and high K⁺ application, which produced only small Ca²⁺ transients in the presence of TTX/bafilomycin.

Secondly, AMPA receptor activation did only weakly evoke Ca²⁺ release from internal stores, but induced Ca²⁺ transients that mainly depended on the presence of Ca²⁺ in the bath solution, indicating Ca²⁺ influx. Besides Ca²⁺ influx through AMPA receptors, an increase in Ca²⁺ due to inhibition or reversal of Na⁺/Ca²⁺ exchanger (NCX) upon the AMPA-evoked Na⁺ increase in astrocytes might contribute to the Ca²⁺ transients triggered by AMPA application, as shown before for Na⁺ increases evoked by GABA transport into olfactory bulb and hippocampal astrocytes^22,54. In olfactory bulb astrocytes, NCX-dependent Ca²⁺ increases were small yet sufficient to trigger Ca²⁺-induced Ca²⁺ release, which was abolished upon store depletion with CPA²². AMPA-evoked Ca²⁺ transients in the present study, in contrast, were only weakly affected by store depletion, indicating that they were not mainly mediated by NCX-dependent Ca²⁺-induced Ca²⁺ release.

Thirdly, our findings were confirmed with acutely dissociated cells excluding indirect effects through neuronal activation. The observation that AMPA receptor currents were sensitive to Naspm further substantiated a Ca²⁺ permeability of the receptors. It should be noted that during our patch-clamp experiments, AMPA receptor-evoked currents were isolated by blocking potassium conductances with Ba²⁺ and quinine, which affect membrane properties of astrocytes^55,56 and suppress Na⁺-dependent block of K⁺ channels^32,33; hence, in the absence of K⁺ channel blockers, AMPA receptor-evoked effects on membrane currents might be more complex than shown in our experiments. Importantly, we have demonstrated that Ca²⁺-permeable AMPA receptors of periglomerular astrocytes are activated through electrical stimulation of ORN axons. ORN axons release glutamate and ATP^18,19,21,57, and electrical stimulation of ORN axons as well as odor stimulation of ORN has been shown to trigger Ca²⁺ signaling in periglomerular astrocytes by activation of mGluR₅, P2Y₁ and A_2A receptors^20,21. However, in the present study we have inhibited these receptors as well as dopamine receptors, which in the olfactory bulb are expressed by many neurons^58,59,60, to reduce potential indirect effects via dopaminergic neurons. The stimulation-induced Ca²⁺ response in periglomerular astrocytes remaining in this blocker cocktail was sensitive to Naspm, indicating that this form of axon-glia interaction activates Ca²⁺-permeable AMPA receptors. Olfactory bulb astrocytes have been shown to mediate neurovascular coupling^20,22,23 as well as release of glutamate, GABA and ATP affecting mitral cells and granule cells^24,61. Accordingly, astrocytes might sense the level of axonal activity by gradual activation of Ca²⁺ influx through their AMPA receptors, thereby modulating gliotransmitter release and adapting local circulation and energy supply to the actual metabolic requirements. In addition, astrocytes are involved in the development of the glomeruli, and AMPA receptor-mediated astrocyte Ca²⁺ signaling might affect neurite growth and synaptogenesis^62,63.

Material and Methods

Animals used for slice preparation

For Ca²⁺ imaging experiments, NMRI, GCaMP6s^fl/fl and GLAST-CreETR2 mice of both genders at postnatal days 14 to 40 were used^64,65. NMRI and GLAST-CreETR2xGCaMP6s^fl/fl mice were raised in the animal facility at the University of Hamburg (Germany). hGFAP-eGFP mice⁶⁶ used for electrophysiology and immunofluorescence staining were obtained from the animal facility at the University of Bonn Medical Center (Germany). All experiments were performed in accordance with EU and local animal welfare guidelines and were approved by the state’s animal welfare committee (GZ G21305/591-00.33; Behörde für Gesundheit und Verbraucherschutz, Hamburg, Germany). Olfactory bulbs were prepared and sliced (VT1200, Leica, Benzheim, Germany) in cooled, carbogen-gassed preparation solution and transferred to carbogen-gassed ACSF at 30 °C for recovery.

Solutions and chemicals

The following solutions were employed (molarity in mM), ACSF: 120 NaCl, 2.5 KCl, 1 NaH₂PO₄ × 2H₂O, 26 NaHCO₃, 2.8 D-(+)-glucose, 1 MgCl₂, 2 CaCl₂; Ca²⁺-free ACSF: 120 NaCl, 2.5 KCl, 1 NaH₂PO₄ × 2H₂O, 26 NaHCO₃, 2.8 D-(+)-glucose, 3 MgCl₂, 0.5 EGTA; preparation solution for Ca²⁺ imaging: 83 NaCl, 1 NaH₂PO₄ × 2H₂O, 26.2 NaHCO₃, 2.5 KCl, 70 Sucrose, 20 D-(+)-glucose, 2.5 MgSO₄ × 7 H₂O; preparation solution for electrophysiology: 87 NaCl, 1.25 NaH₂PO₄ × 2H₂O, 25 NaHCO₃, 2.5 KCl, 7 MgCl₂ × 6 H₂O, 0.5 CaCl₂ × 6 H₂O, 60 Sucrose, 25 D-(+)-glucose (325 mOsm). ACSF, preparation solutions and Ca²⁺-free ACSF were continuously gassed with carbogen (95% O₂, 5% CO₂) to buffer the pH at 7.4 and to supply oxygen. Patch clamp-recording of isolated cells was performed in bath solution containing (in mM): 150 NaCl, 5 KCl, 2 MgCl₂ × 6 H₂O, 2 CaCl₂ × 6 H₂O, 10 D-(+)-glucose, 10 HEPES, pH 7.4, gassed with oxygen. The compounds amino-3-hydroxy-5-methyl-4-isoxazolephosphonic acid (AMPA), cyclothiazide, (E)-ethyl 1,1a,7,7a-tetrahydro-7-(hydroxyimino)cyclopropa[b]chromene-1a-carboxylate (CPCCOEt), 2-methyl-6-(phenylethynyl)pyridine (MPEP), 2′-deoxy-N6-methyladenosine 3′,5′-bisphosphate (MRS2179), 2-(2-furanyl)-7-[3-(4-methoxyphenyl)propyl]-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin-5-amine (SCH442416) and (R)-(+)-7-chloro-8-hydroxy-3-methyl-1-phenyl-2,3,4,5-tetrahydro-1H-3-benzazepine (SCH23390) were obtained from Abcam (Cambridge, United Kingdom). Adenosine 5′-triphosphate (ATP), kainate, N-methyl-D-aspartic acid (NMDA) and papain was purchased from Sigma Aldrich (Taufkirchen, Germany), GYKI 53655 from Tocris (Bristol, UK). Bafilomycin A1 and CPA were acquired from Enzo Life Sciences (Lörrach, Germany). The reagents D-APV, Naspm and TTX were received from Alomone labs (Jerusalem, Israel). All reagents were stored as stock solutions corresponding to the manufacturer’s instructions and added to ACSF directly before the experiment.

Ca²⁺ imaging

Tissue slices were placed in a recording chamber and fixed with a platinum frame fitted with nylon strings. Slices were incubated with the membrane-permeable form of the Ca²⁺ indicator Fluo-4 (Fluo-4-AM; 2 μM in ACSF) made from a 4-mM stock solution (dissolved in DMSO and 20% pluronic acid) for 40 min. In most of the experiments, bafilomycin A1 (10 μM) was added to the Fluo-4-AM solution. In some experiments, GCaMP6s fluorescence was used as an indicator of Ca²⁺ concentration. Changes in intracellular Ca²⁺ levels in periglomerular astrocytes were recorded by confocal microscopy (C1 Eclipse, Nikon, Düsseldorf, Germany). An excitation laser wavelength of 488 nm and a frame rate of 0.3–0.5 fps were used. Drugs were administered via the perfusion system except for AMPA. The application solution containing AMPA was applied directly into the perfusion stream in the bath with a custom made application system to allow for a semi-fast application (within 3–5 s). The flow rate of the application system equaled the flow rate of the perfusion system, resulting in a 1:2 dilution of the agonist concentration as adjusted in the application solution.

For electrical stimulation of axons in tissue slices, a glass pipette with a resistance of 2–2.5 MΩ filled with ACSF was used. The pipette was placed on the surface of the olfactory nerve layer (ONL) comprising axons of olfactory receptor neurons and 250-μA stimuli were applied for 2 s at 10 Hz.

Isolation of periglomerular astrocytes

For tissue preparation, 300 μm thick sagittal slices from the olfactory bulb were prepared from hGFAP-eGFP mice (postnatal day 8–14) in ice cold, carbogen (95% O₂/5% CO₂) gassed ACSF supplemented with sucrose (preparation solution) using a vibratome (VT1200S, Leica, Nussloch, Germany). Slices were transferred for 15 min to warm (35 °C) preparation solution and then to standard ACSF (room temperature, 1 h). Cells were isolated using an enzymatic/mechanical approach¹³. Slices were incubated in ACSF supplemented with papain (1.5 mg/ml; 24 U/ml) (Sigma, Taufkirchen, Germany) and L-cysteine (0.35 mg/ml) (Sigma) (10 min) and continuously bubbled with carbogen. Subsequently, slices were transferred to the recording solution and the glomerular layer was dissected under a stereomicroscope (KL200, Zeiss, Jena, Germany). Cells were isolated in the recording chamber with tungsten needles and Pasteur pipettes, and allowed to settle for 15 min before analysis. Periglomerular astrocytes were identified by their green fluorescence and morphology. For Ca²⁺ imaging experiments, astrocytes were isolated from NMRI mice, seeded on concanavalin A-coated cover slipes, loaded with Fluo-4-AM (2 μM in ACSF) for 30–45 min and imaged as discribed for brain slices.

Electrophysiological recordings

Experiments on isolated cells employed a customized concentration clamp device connected to an EPC-7 amplifier and TIDA software (Heka Lambrecht, Germany) as described elsewhere¹³. Astrocytes were visualized with an inverted microscope (Axiovert 135, Zeiss) equipped with DIC and epifluorescence. Pipettes were manufactured from borosilicate glass (2–4 MΩ; Science Products, Hofheim, Germany) and filled with a solution containing (in mM): 130 KCl, 0.5 CaCl₂, 2 MgCl₂, 5 BAPTA, 10 HEPES, and 3 Na₂-ATP, 0.05 spermine, pH 7.25. Currents were sampled at 0.1 to 30 kHz and filtered at 3 or 10 kHz. Holding potential was −70 mV. Input and series resistance were continuously checked by applying 10 mV test pulses. The liquid junction potential was not corrected for. Recordings were performed at room temperature. To separate the AMPA receptor conductance from simultaneously occurring changes in K⁺ conductance, drug application to isolated cells was performed in HEPES-buffered recording solution, supplemented with the K⁺ channel blockers quinine (100 μM) and BaCl₂ (100 μM)³³.

Data analysis

To analyze changes of the Ca²⁺ level in astrocytes, cell somata were marked as regions of interests (ROIs) using EZC1 Viewer software (Nikon). Cells located in the glomerular layer that showed a Ca²⁺ response to ATP were identified as periglomerular astrocytes²¹. To analyze changes of Ca²⁺ levels over time, Fluo-4 and GCaMP6s fluorescence intensity (F), respectively, was recorded throughout the experiment and normalized to the basal fluorescence intensity in absence of stimuli (F₀). Changes in Ca²⁺ are given by ΔF/F₀. All values are given as mean values ± standard error of the mean. Data for every set of experiment was acquired from at least three different animals. The assessment of statistical significance by comparing two means was done by Student’s t-test or, if applicable, one-way ANOVA with Fisher’s post-hoc test at an error probability p (*p < 0.05; **p < 0.01; ***p < 0.001).

Immunohistology

Immunohistological staining was performed on 100-μm thick sagittal slices of olfactory bulbs of NMRI and hGFAP-eGFP mice. For anti-GluA staining, cerebella of the same animals were used as a positive control to verify specific antibody staining. After dissection, olfactory bulbs and cerebella were stored and refrigerated in paraformaldehyde solution (PFA, 4%) in phosphate buffered solution (PBS) containing (in mM): 130 NaCl, 7 Na₂HPO₄, 3 NaH₂PO₄. Slices were cut with a vibratome (VT1000, Leica) and incubated with the primary antibodies anti-GluA1 (guinea pig; 1:200; Alomone Labs), anti-GluA2 (rabbit; 1:200; Millipore, Darmstadt, Germany), anti-GluA4 (rabbit; 1:200; Millipore), anti-GFAP (rabbit, 1:000, Dako, Hamburg, Germany), anti-GFAP (chicken; 1:500; Abcam) and anti-GFP (chicken, 1:500, SySy, Göttingen, Germany). Antibodies were diluted in 1% NGS, 0.05% TritonX100 in PBS. Subsequently, slices were incubated with secondary antibodies over night at room temperature. Secondary antibodies (1:1000 in PBS) used were: goat anti-rabbit Alexa Fluor 488 (Invitrogen Thermo Fisher, Darmstadt, Germany), goat anti-rabbit Alexa Fluor 555 (Invitrogen Thermo Fisher), goat anti-chicken Alexa Fluor 555 and goat anti-chicken Alexa 488 (Abcam), CF488A donkey anti-guinea pig (Sigma-Aldrich). Hoechst 33342 (5 μM; Molecular Probes, Eugine, USA) was used for nuclear staining. Stacks of confocal images were acquired (C1 Eclipse, Nikon) using a 40x/NA 1.3 oil immersion lens. The axial step size was 150 nm. Image stacks of GluA staining were deconvolved using Huygen’s software (SVI, Hilversum, Netherlands). Projections were made using Image J (NIH, Bethesda, USA) and adjusted for brightness and contrast.