SynaptopHluorin fluorescence correlates with presynaptic electrical activity
Odor-evoked changes in spH fluorescence are linked to presynaptic activity (Bozza et al., 2004
; Wachowiak et al., 2005
), but the exact relation of spH signal to electrical activity in vivo has not been quantified. Therefore, we recorded multi-unit activity of olfactory sensory neurons and spH fluorescence from glomeruli in mice that express spH under control of the olfactory marker protein (OMP) promoter (Bozza et al., 2004
). We recorded fluorescence with a CCD camera through a cranial window (). Mice were stimulated with different concentrations of methyl tiglate (0.1 – 6.4 %). To avoid habituation, we presented 5 different odors between each concentration step. Each stimulus was preceded by pure air. We identified activated glomeruli by their fractional fluorescence change compared to air (0.61 ± 0.09 % ΔF/F; n = 18 glomeruli from 5 mice). Electrodes were then placed into responsive glomeruli (). We observed a linear monotonic relation between spH fluorescence and the integrated olfactory nerve spike rate (; r = 0.90, p < 0.001, Pearson correlation), indicating that spH fluorescence can be used as a marker of presynaptic activity over a wide dynamic range of physiological stimuli.
spH fluorescence correlates with presynaptic electric activity
To demonstrate that spH fluorescence indicates transmitter release, and not spike activity, we applied baclofen to the recording region by local microinjection (10 mM, n = 2 mice). Baclofen decreases glomerular glutamate release by activating presynaptic γ-aminobutyrate-B (GABAB
) receptors (Wachowiak et al., 2005
). spH fluorescence was strongly reduced, while presynaptic spike rate remained unchanged (), consistent with the origins of the recorded multi-unit activity in the olfactory nerve layer, and that of spH signal in presynaptic terminals.
Simultaneous multiphoton imaging of CBF and presynaptic activity following physiological stimulation
We used multiphoton microscopy to monitor spH fluorescence and CBF simultaneously in glomeruli in vivo (). We visualized blood vessels by tail vein injections of Texas Red dextran (). Texas Red and spH could be excited at the same wavelength with minimal bleed-through in emission (Video S1
). Vessels in the glomerular layer were capillaries () as previously reported (Chaigneau et al., 2003
), while surface arteries and arterioles were located in the olfactory nerve layer ( and see below). Higher magnifications of the glomerular layer () revealed erythrocytes as dark objects moving in the fluorescently labeled plasma (, Video S1
). Their velocity and flux were determined by line scans along the central axis of glomerular capillaries (Dirnagl et al., 1992
; Kleinfeld et al., 1998
). By extending the line scans beyond the capillary length and into the glomerular tissue, we recorded changes in spH fluorescence and CBF simultaneously at high temporal and spatial resolution (). Upon odor stimulation, velocity and flux increased ().
Simultaneous multiphoton imaging of presynaptic activity and CBF in vivo
Local glomerular CBF is correlated with presynaptic glutamate release
We used 25 different odorants (Table S1
) to investigate the relation between spH intensity and functional hyperemia. Odors and recording regions were selected to achieve a sparse glomerular activation pattern (~1 activated glomeruli per field of view) to avoid possible overlap of neighboring co-activated glomeruli and to study responses of single glomeruli.
Stimulation led to an increase in spH fluorescence (5.8 ± 0.4 %; n = 103 glomeruli from 21 mice), with a large range from 0.3 % to 16.9 %. Baseline CBF parameters were obtained during the 10 s presentation of fresh air (velocity, 0.65 ± 0.11 mm/s; flux, 64.1 ± 9.8 s−1; n = 152 capillaries from 21 mice). An increase of velocity and flux was observed in 94 % of activated glomeruli. The onset of functional hyperemia, measured as the rise of flux to half-maximal amplitude, occurred later than the onset of the simultaneously recorded spH signal (difference of 2.2 ± 0.3 s, p < 0.05, paired t-test).
Following stimulation, velocity increased by 25.4 ± 3.6 % and flux by 34.5 ± 4.1 % (Video S3
). The changes in velocity and flux showed a wide dynamic range (velocity 5.4 % to 71.3 %, flux 8.4 % to 76.9 %) and were correlated with the amplitude of spH increase (). In addition, spH and CBF changes exhibited similar habituation when the same stimulus was intentionally presented in short temporal progression (). The correlation between spH fluorescence and CBF changes was significant (velocity, r = 0.90; flux, r = 0.81, p < 0.001, Pearson correlation), and the data were well fitted by a linear regression ().
Presynaptic glutamate release and CBF are highly correlated
Local postsynaptic blockade does not affect functional hyperemia in glomeruli
Since local CBF changes were correlated with presynaptic activity (and therefore, release of glutamate), we investigated the involvement of postsynaptic N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptors, which attenuate functional hyperemia in the neocortex and cerebellum (Lauritzen, 2005
). To locally block these receptors, we injected antagonists into glomeruli. Injection pipettes were co-loaded with Texas Red to confirm localized injections into glomeruli of interest (Fig. S1
To confirm blockade of ionotropic glutamate receptors, we recorded calcium changes in juxtaglomerular cells and mitral/tufted (M/T) cell dendrites following odor stimulation. Cells were labeled with the calcium indicator X-Rhod-1 acetoxymethyl ester (AM) by multicell bolus loading () (Stosiek et al., 2003
). In responding glomeruli, we detected calcium transients in juxtaglomerular cells in all cases (; ΔF/F 23 ± 6 %, n = 81 cells from 5 mice). Calcium increased 0.9 ± 0.2 s after the onset of the spH increase. Local injection of the NMDA receptor blocker APV (50 mM) and the AMPA receptor inhibitor 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX; 5 mM) abolished responses in 94 % of cells (). A fraction of these cells likely represent astrocytes, which are sensitive to mGluR blockade (see below). Moreover, the spH increase following odor stimulation was significantly stronger after NMDA/AMPA blockade, consistent with presynaptic disinhibition () (McGann et al., 2005
). This presynaptic augmentation was abolished when baclofen (100 μM) was coinjected with APV/CNQX ().
Local postsynaptic activity is not required for functional hyperemia
Although activity of periglomerular cells was abolished by injection of APV/CNQX, it remained possible that the dendrites of the principal M/T cells retained some responsiveness. We therefore labeled dendrites of M/T cells with X-Rhod-1 AM using a modified loading protocol (see Experimental Procedures). Optical sections starting from the glomerular region going towards the mitral cell body layer were used to identify apical dendrites of M/T cells (). We then measured odor-evoked calcium responses in apical dendrites (; peak ΔF/F 18 ± 3 %; n = 4 mice). These responses were fully blocked by local injection of APV and CNQX (), again confirming blockade of postsynaptic activity by this treatment.
Although postsynaptic activity was blocked by local injection of APV/CNQX, odor-evoked hyperemia was not reduced (). In fact, it was slightly but significantly stronger, consistent with more glutamate released presynaptically and, accordingly, a stronger CBF response (; n = 3 mice). When baclofen (100 μM) was injected together with APV/CNQX, the fractional spH increase and functional hyperemia were not different compared to control (; n = 2 mice).
Astrocytes bridge glomerular synapses with blood vessels
Since release, but not local ionotropic activity of glutamate, was correlated with functional hyperemia, we next investigated the extrasynaptic contributions of glutamate. Because astrocytes detect glutamate released at glomerular synapses (De Saint Jan and Westbrook, 2005
), we speculated that they act as an intermediary between glutamate and CBF changes. Staining for glial acidic fibrillary protein (GFAP) in fixed olfactory bulb slices demonstrated that each astrocyte sends its processes into a single glomerulus (). Double immunostaining for GFAP and the vascular marker laminin B2/γ1 revealed that within glomeruli, astrocytic processes reach into the glomerular core and contact capillaries (). Astrocytes contacted both glomerular capillaries as well as larger upstream vessels feeding into glomeruli (, inset). Lower magnification images revealed that the complete vascular tree – i.e. large surface vessels, medium-sized vessels penetrating the olfactory nerve layer, and glomerular capillaries – is covered by astrocytes (). Furthermore, staining for aquaporin-4, a marker for astrocytic endfeet (Simard et al., 2003
), revealed a dense astrocytic network around blood vessels and throughout spH-negative portions of glomeruli ().
Astrocytes couple synapses with the vasculature
Astrocytes throughout the vascular network show odor-selective calcium elevations
To probe astrocytic activity in vivo
, we labeled the olfactory bulb of mice expressing GFP under the GFAP promotor with X-Rhod-1 AM by multicell bolus loading. Olfactory astrocytes are strongly labeled with GFP in these mice (De Saint Jan and Westbrook, 2005
). Astrocytes exhibited calcium responses to odor stimulation () that commenced shortly after stimulus onset (0.9 ± 0.3 s). Glomerular astrocytes responded specifically to only one or very few odors (), indicating that they are stimulated by glutamate released in activated glomeruli.
Astrocytic calcium responses in glomeruli are conveyed upstream and are associated with arteriolar dilation
To test whether periarteriolar astrocytes in the olfactory nerve layer also respond to odors, we labeled astrocytes by topical application of X-Rhod-1 AM (Hirase et al., 2004
). Blood vessels were visualized by tail vein injections of fluorescein isothyocyanate (FITC) dextran. Arterioles were distinguished from venules by their thicker wall and by their direction of flow away from surface vessels. Similar to glomerular astrocytes, astrocytic endfeet located around upstream arterioles of the same vascular network also responded to odor stimulation (; ΔF/F 4.3 ± 0.4 %). These calcium elevations appeared shortly after stimulus onset (1.0 ± 0.2 s). Because astrocytic endfeet moved rhythmically with heartbeat- and breathing-related movements, we could obtain successful recordings only in a subset of astrocytes (n = 11 endfeet from 6 mice).
Odor-evoked calcium responses in astrocytic endfeet are associated with arteriolar dilation
The main mechanism regulating CBF in capillaries is a change in the diameter of upstream arterioles. We hypothesized that astrocytic endfeet around penetrating arterioles would receive odor-evoked signals from glomerular astrocytes and convey these signals onto arterioles. To test this possibility, we measured changes in arteriolar cross-sectional area by labeling the vasculature with FITC dextran. We found that odor stimulation resulted in a significant increase in the cross-sectional area of penetrating arterioles between 13.4 % and 35.1 % (mean: 21.9 ± 4.2 %), corresponding to a flow increase between 29 % and 81 % (assuming Poiseuille’s law). Baseline fluctuations in cross-sectional area without stimulation were significantly lower (3.4 ± 1.1 %).
We then combined calcium imaging of astrocytes in GFAP-GFP mice with arteriolar cross-sectional area recordings. We found that odor-evoked calcium elevations in periarteriolar astrocytic endfeet were accompanied by increases in arteriolar cross-sectional area (). The two signals were strongly correlated temporally and spatially: Invariably, calcium elevations in astrocytic endfeet preceded or coincided with arteriolar dilation, and non-reacting arterioles showed no calcium changes in surrounding endfeet. Small baseline fluctuations of arteriolar area were not associated with astrocytic calcium changes.
Glutamate mediates functional hyperemia via astrocytic mGluR5
We then sought to determine the intracellular pathways responsible for the vascular changes. These experiments were carried out in glomerular capillaries rather than in arterioles to simultaneously monitor spH responses and ensure that they were not altered in non-specific ways. Arterioles are above glomeruli and therefore spH cannot be recorded in those optical sections.
We tested whether mGluRs are involved in functional hyperemia following sensory stimulation. Topical application of the group I/II mGluR antagonist (S)-α-methyl-4-carboxyphenylglycine (MCPG, 50 mM) significantly reduced the odor-evoked increase of velocity by 48 ± 3 % and flux by 52 ± 4 % (n = 13 glomeruli from 4 mice; p < 0.05, paired t-test). Since topically applied drugs may not adequately penetrate brain tissue, we also microinjected MCPG into the bulb, which also reduced velocity and flux responses at two different concentrations (1 mM and 50 mM; ). Importantly, spH fluorescence remained unchanged ().
Astrocytes mediate functional hyperemia through mGluR5 and glutamate uptake
Several mGluR subtypes are also expressed by M/T cells (Ennis et al., 2006
), and astrocytes respond to currents generated by mGluR located on these cells (De Saint Jan and Westbrook, 2005
). Therefore, we performed additional experiments with 6-methyl-2-(phenylethynyl)-pyridine (MPEP), a selective antagonist of mGluR5, which in the glomerular layer is exclusively expressed by astrocytes (van den Pol, 1995
). MPEP (100 μM) injected into the bulb led to a significant reduction (~40%) of functional hyperemia (). Again, spH fluorescence was not altered (, paired t-test).
To ascertain that sufficient concentrations of mGluR blockers had been applied, we recorded calcium signals from glomerular astrocytes in GFAP-GFP mice (). MCPG (1 mM) and MPEP (100 μM) both suppressed odor-evoked calcium responses in astrocytes (). MCPG reduced responses in most juxtaglomerular cells, while MPEP was specific for astrocytes ().
Astrocytic glutamate uptake represents an additional pathway of functional hyperemia
Because a significant fraction of the CBF response remained unresponsive to mGluR inhibition, we explored the involvement of additional astrocytic signaling pathways. Another major pathway for glutamatergic actions on astrocytes involves uptake through amino acid transporters (Marcaggi and Attwell, 2004
). Glutamate uptake has been implicated in the initiation of astrocytic glucose utilization (Voutsinos-Porche et al., 2003
), and in the generation of intrinsic optical signals in the olfactory bulb (Gurden et al., 2006
To investigate whether glutamate uptake into astrocytes represents an additional neurovascular signaling pathway, we first applied the broad-spectrum glutamate transporter inhibitor DL-threo-β-Benzyloxyaspartic acid (TBOA) topically to the olfactory bulb (10 mM; n = 5 mice). This treatment reduced odor-evoked velocity increase by 40 ± 3 % and flux increase by 35 ± 3 % (; n = 18 glomeruli from 6 mice; p < 0.05, paired t-test). A similar effect was observed when TBOA was injected into the glomerular region (100 μM, velocity 39 ± 3 %, flux 35 ± 5 % reduction from control; n = 8 glomeruli from 3 mice). SpH responses remained similar to control ().
TBOA inhibits both astrocytic and neuronal glutamate transporters, both of which are expressed in glomeruli (Utsumi et al., 2001
). To achieve higher selectivity for astrocytic glutamate uptake, we applied dihydrokainate (400 μM) by local microinjection. Dihydrokainate selectively inhibits the astrocytic glutamate transporter GLT-1, which is also expressed in glomeruli (Utsumi et al., 2001
). Dihydrokainate reduced the odor-evoked CBF response, while the spH increase remained unchanged (; n = 15 glomeruli from 5 mice; p < 0.05, paired t-test). A similar reduction was also observed at a higher concentration (1 mM; velocity, 32 ± 4 %; flux, 26 ± 4 %; n = 3 mice).
Inhibition of glutamate uptake into astrocytes increases the extracellular concentration and half-life of glutamate, which could alter its action on glutamate receptors. We indeed observed this effect in initial experiments, which showed stronger postsynaptic responses after local TBOA injection (data not shown). To control for potentially higher ionotropic and metabotropic receptor activity after glutamate transport blockade, we blocked glutamate receptors by combined local injection of APV (50 mM), CNQX (5 mM), and MCPG (50 mM), and presynaptic GABAB receptors with baclofen (100 μM). This treatment reduced odor-evoked CBF responses (). However, functional hyperemia decreased even further after subsequent injection of dihydrokainate (1 mM; ; n = 7 glomeruli from 3 mice) or TBOA (1 mM; ; n = 6 glomeruli from 3 mice), indicating that glutamate uptake contributes to functional hyperemia independently of ionotropic and metabotropic glutamate receptors, as well as presynaptic GABAB receptors. The effects of TBOA and MPEP were found for all odors without apparent spatial or odor-specific differences.
Functional hyperemia mediated by mGluR, but not by glutamate uptake, depends on cyclooxygenase
As noted above, the calcium increase in glomerular astrocytes was dependent on mGluR5 (). However, in accordance with other reports (Bernardinelli et al., 2004
), we observed no significant changes of the astrocytic calcium response when TBOA or dihydrokainate were injected locally into glomeruli of GFAP-GFP mice (TBOA, 19.3 ± 3.2 vs. 22.5 ± 3.4 ΔF/F %; DHK, 17.1 ± 4.2 vs. 19.1 ± 3.7 % ΔF/F %; paired t-test; n = 26 astrocytes from 4 mice). There was a slight tendency for the calcium responses to increase, but this effect did not reach statistical significance.
Direct activation of astrocytes induces vasodilation by COX1 activation and prostaglandin synthesis (Zonta et al., 2003a
; Takano et al., 2006
). We asked whether the pathways mediated by mGluR and glutamate uptake are also COX1-dependent. Immunohistochemistry of bulb slices from GFAP-GFP mice revealed that COX1 is strongly and exclusively expressed by astrocytes in the glomerular layer (), indicating that COX1 inhibitors can be used to selectively target glomerular astrocytes. In contrast, COX2 expression was sparse and mostly observed in neurons below the glomerular layer (Figure S2
Functional hyperemia mediated by mGluR and glutamate uptake is controlled by separate pathways
The COX1 inhibitor SC-560, applied by local microinjection (500 μM), reduced odor-evoked velocity and flux increase by 34 ± 6 % and by 36 ± 7 %, respectively (; n = 12 glomeruli from 4 mice). No significant further reduction was seen after subsequent local injection of the mGluR5 antagonist MPEP (), indicating that COX1 activation occurs downstream of mGluR5 activation. Moreover, the exclusive expression of both proteins in astrocytes indicates that they are part of a common astrocytic signaling pathway. In contrast, in mice treated with SC-560 in a separate group (n = 10 glomeruli from 4 mice), local injection of TBOA reduced functional hyperemia even further (). Similarly, application of TBOA resulted in an additional reduction of functional hyperemia when applied after MPEP in the same animal (MPEP: velocity, 42 ± 3 %, flux 41 ± 6 %; TBOA: velocity 59 ± 5 %, flux 61 ± 3 %, reduction of hyperemia relative to control; n = 18 glomeruli from 3 mice; p < 0.05, Repeated Measures ANOVA followed by Tukey Test).
Cerebrovascular reactivity following local drug application
Finally, we ascertained that the observed changes of the CBF response were due to the specific modulation of functional hyperemia induced by neural activity. Baseline CBF values are reported in . Drugs were co-injected with FITC (Fig. S3
). To test vascular reactivity, we bolus-injected the carbonic anhydrase inhibitor acetazolamide (14 mg/kg i.v.), which induces vasodilation independently of neural activity by increasing tissue pCO2
. Acetazolamide induced an increase in velocity and flux in glomerular capillaries (Fig. S3
). No differences were observed after injections of MPEP, DHK, TBOA, and APV/CNQX, but SC-560 reduced pCO2
reactivity (n = 2 mice for each; Fig. S3
) as previously reported (Niwa et al., 2001
). However, this reduced pCO2 reactivity is unlikely to be a confounding factor, because SC-560 interacted differently with TBOA and MPEP in terms of its effect on functional hyperemia.