Characterization of FLII81E - 1μ sensor loading in brain slices
Cortical slices were loaded with FLII81E - 1μ using an interface-incubation loading technique as described above. Imaging of Venus and CFP fluorescence demonstrated the presence of FRET sensor in the loaded slice. The fluorescence ratio was stable over time (, < 5% decrease in fluorescence ratio over 25 minutes, 200 ms of fluorescence illumination every 10 seconds) although CFP and Venus fluorescence decreased with time, presumably due to a combination of protein degradation, and wash-out from the slice (, values reflect dark-noise adjusted raw fluorescence values). To isolate the effects of sensor wash-out from sensor photobleaching, we imaged slices loaded with the FRET sensor for 20-25 minutes using either a short or long fluorescence exposure time (see methods). During long exposures FRET sensor signal should decay as a function of sensor washout and sensor bleaching/degradation. During short exposure times, FRET sensor signal decay should be much more dependent on sensor washout, as minimal bleaching should occur. During long exposures FRET sensor fluorescence ratio and individual fluorescence channels decreased more rapidly. By subtracting the short exposure decay curve from the long exposure decay curve we were able to compute the decay constants for both the bleaching and washout of FRET sensor from cortical brain slices, both of which were over 30 minutes (data not shown).
FLII81E - 1μ glutamate sensor loading in cortical slices
We also estimated sensor loading and washout (unloading) using quantitative western blotting. Brain slices initially contained an estimated 70μM FRET sensor immediately after loading, but before any superfusion (data not shown), and following 20 minutes of superfusion the FRET sensor concentration decreased to approximately 20μM, suggesting that washout occurs over several tens of minutes. Slices used for imaging experiments presumably have a lower initial FRET sensor concentration, due to the washout that occurs during the period (~5 min) in which the slice and electrodes were positioned prior to each recording.
Tissue cross sections obtained from slices that were fixed and sectioned following recording demonstrated that the glutamate FRET sensor fluorescence was present throughout the depth of the slice indicating that the loading procedure resulted in relatively uniform FRET sensor protein penetration (). Slices not loaded with FLII81E - 1μ protein showed no Venus fluorescence (). Slice health was confirmed using DAPI staining to show intact nuclei (). A different set of slices was sectioned horizontally to address the uniformity of FLII81E - 1μ loading in the plane of the slice. Glutamate FRET sensor signal was spatially uniform within each individual section and sections from the top 50 μm of tissue (), as well as from 150 - 200 μm (), and 300-350 μm () deep into the tissue all had similar Venus fluorescence. This result indicates that there were no significant regional differences in loading of the slice. The penetration of FLII81E - 10μ showed a similar pattern (data not shown).
FLII81E - 1μ detects evoked glutamate release in cortical brain slices
The FLIPE - 600n glutamate sensor was the first FRET sensor characterized. FLIPE - 600n successfully loaded into slices and was able to resolve glutamate transients under disinhibited but not under control conditions (data not shown). To confirm that the sensor response was due to neurotransmitter release 1 μM TTX was used to block action potentials. Addition of 1 μM TTX blocked the glutamate transient seen in disinhibited slices (data not shown), indicating that the glutamate signals seen are due to evoked glutamate release.
FLII81E - 1μ, a sensor with a 1μM affinity for glutamate, was loaded into slices and evoked cortical field potentials were recorded simultaneously with image capture. FLII81E sensors have a lower affinity than FLIPE sensors, but utilize a different protein design optimized for the greatest glutamate-induced FRET ratio change (up to 3X larger than FLIPE). ROI analysis was performed on the area of the highest glutamate signal; the same ROI was used for all manipulations within each slice. Under control conditions, electrical stimulation of the white matter caused a transient decrease in FLII81E - 1μ fluorescence ratio (, -2.65 ± 0.3% peak Δ FY/FC, n=23). The response peaked 63.3 ± 5.8 ms following stimulation and had a half width of 237.7 ± 9.2 ms. The signal was localized near the stimulation electrode, approximately in cortical layers 4 and 5 (). Under control conditions cortical field potentials were small (<0.2 mV) and brief (<50 ms) (). Excitability was then increased by local perfusion of 10 μM GABAzine, a GABAA receptor antagonist.
High-affinity FLII81E - 1μ glutamate sensor imaging
A larger and more prolonged decrease in FLII81E - 1μfluorescence ratio was seen when GABAA-mediated inhibition was blocked (-19.1 ± 0.5% peak ΔFY/FC, 416.1 ± 10.3 ms half width, n=25, p<0.01 compared to control). This glutamate transient reached its maximum later than control (, 78.5 ± 2.9 ms after stimulation, p<0.05, n=25). The evoked glutamate transient seen in the presence of GABAzine generally manifested in a column-like band of cortex near the site of stimulation and then spread to adjacent cortical areas, indicating the importance of inhibitory interneurons in shaping the functional activation of the cortical network. Blockade of GABAA receptors induced a second, later phase of the field potential (indicative of extended cortical network activation), which increased in both amplitude (≈ 0.5 mV) and in duration (≈ 1.5 sec) ().
Inhibition of the EAAT family of glutamate transporters by TBOA (25 μM) caused a decrease in baseline FLII81E - 1μ ratio (7.7 ± 0.4 % decrease in FRET sensor ratio after 10 minutes of exposure, n=6), indicating approximately 100 nM accumulation of extracellular glutamate (TBOA did not alter the properties of the FRET sensor in cuvette studies, data not shown). In the presence of TBOA and GABAzine, electrical stimulation produced a prolonged decrease in FLII81E - 1μ fluorescence ratio (877.5 ± 59.7 ms half width, n=25, p<0.01 compared to GABAzine alone) which peaked later (126.5 ± 9.9 ms after stimulation, n=25, p<0.01 compared to GABAzine alone) although the peak amplitude of the transient was relatively unchanged (-19.3 ± 0.8 % decrease in FLII81E - 1μ fluorescence ratio, n=25, n.s.). Under these conditions the FLII81E - 1μ signal spread over a larger area of cortex, although the amplitude of glutamate transients was relatively small in layers 4 and 5 compared to layers 2/3 and 6. Blockade of glutamate transporters also prolonged the field potential duration (> 2 sec) and increased the time to the peak of both the early (, inset) and the late phases of the field potential () although the effects on amplitude varied between slices. The individual CFP and Venus signals for all manipulations are shown () and indicate that changes in FY/FC ratio are caused by changes in FRET efficiency (i.e. increased CFP and decreased Venus signal). Using this protocol all slices had a detectable change in FRET sensor signal and each stimulation caused a similar FRET sensor response (5/5 trials per slice per manipulation), although there was variability in the amplitude and activation pattern between slices.
Evoked glutamate release is not detectable using FLII81E - 10μ
In order to estimate the amount of glutamate released in the slice, we repeated the experiments performed with FLII81E - 1μ, using FLII81E - 10μ a glutamate sensor with a 10-fold lower affinity. The cortical field potentials recorded using FLII81E - 10μ were similar to previous experiments, as were the effects of GABAzine and TBOA (). No change in FLII81E - 10μ fluorescence ratio was seen for any of the manipulations tested (). In order to confirm that this FRET sensor was functional, we locally perfused 10 mM glutamate onto slices loaded with FLII81E - 10μ (), which caused a decrease in the Venus/CFP fluorescence ratio. Thus, although FLII81E - 10μM did not detect evoked glutamate release under the conditions tested it was fully functional in its ability to detect glutamate at higher concentrations.
Low-affinity FLII81E - 1μ glutamate sensor imaging
Glutamate transients and cortical field potentials show parallel increase in amplitude and duration during disinhibition
We next examined the changes in excitability and glutamate release that occurred during the wash-in of GABAzine. Soon after the local perfusion was begun (within 30 seconds) the first FLII81E - 1μ images captured showed a detectable, but relatively small and localized glutamate transient (). The simultaneously recorded evoked cortical field was also relatively small and brief (). As excitability progressively increased with prolonged exposure to GABAzine, glutamate transients grew larger in amplitude and duration. Interestingly, as the glutamate transients grew the spatial spread of glutamate FRET sensor signal grew in concert, moving to areas distant from the stimulation electrode. In many trials, cortical layer 4 appeared to have a smaller change in FRET sensor fluorescence ratio compared to layers 2/3, 5 and 6 (). Similar to glutamate transients, cortical field potentials grew progressively larger in amplitude and longer in duration during the wash-in of GABAzine (). Based on these results, changes in field potentials and glutamate transients were correlated. It appears that as inhibition was progressively blocked more glutamate was released with each stimulation and a larger network of neurons became activated ().
FLII81E - 1μ glutamate sensor imaging of a cortical slice during the wash-in of a GABAA receptor antagonist
Spatiotemporal properties FLII81E - 1μ glutamate FRET sensor during full field imaging
A major goal of FLII81E - 1μ glutamate imaging was to increase the spatial and temporal resolution of glutamate detection in slices. As a proof of principle of increased resolution individual images were analyzed to confirm that regional and temporal differences were resolvable. We focused on images captured in the presence of GABAzine or GABAzine + TBOA in order to see large changes in FLII81E - 1μ fluorescence ratio. In many instances, images captured less than 30 ms following stimulation had sub-maximal changes in FLII81E - 1μ fluorescence ratio. These images reflected the earliest phase of glutamate transients, before activation of the entire cortical network has reached its maximum. During these early time-points, regional activation of the cortex was often seen (). In the presence of GABAzine, if an early phase response was captured (early phase activation captured in 12/23 trials) it occurred on average 17.7 ± 5.4 ms after stimulation. In the presence of TBOA, however, early phase responses (early phase activation captured in 12/25 trials) occurred later - 51.1 ± 8.8 ms after stimulation (p<0.01 compared to GABAzine). These results suggests a slower onset or later peak of the FLII81E - 1μ signal in the absence of functional glutamate reuptake, though our ability to resolve temporal changes on this time scale is limited by the relatively slow (≈17 Hz) image capture rate. In one particularly interesting series of three exposures, the inherent jitter in the image acquisition time allowed us to sample a series of time-points close the stimulation time (). When the image capture occurred very close to the time of stimulation (<15 ms, top) the glutamate transient was small. Images captured slightly later (20 ms after stimulation, middle, and 36 ms after stimulation, bottom) had progressively larger amplitude changes in FLII81E - 1μ fluorescence ratio. Although this was a fortuitous example, this result demonstrates the ability of FLII81E - 1μ glutamate sensor imaging to discriminate small changes in glutamate transients on a tens-of-millisecond time scale during full-field image acquisition.
Spatioltemporal properties FLII81E - 1μ glutamate sensor during full field imaging
High-speed (50 Hz) line scan imaging of FLII81E - 1μ glutamate sensor imaging
In order in improve the time-resolution of glutamate FRET sensor imaging, a series of experiments were performed using line-scan image acquisition and simultaneous field recordings. 32 pixel line-scans were collected and binned vertically to decrease image processing time (, dotted area). This allowed image acquisition up to 50 Hz and enabled us to consistently resolve the temporal properties of glutamate transients on a tens-of-millisecond time-scale. Line-scan imaging and ROI analysis of deep (4-6) cortical layers revealed that electrical stimulation caused a transient decrease in FLII81E - 1μ fluorescence ratio under control conditions (, -0.7 ± 0.1 % peak Δ FY/FC, 44.4 ± 6.8 ms peak time, 146.0 ± 10.6 ms half-width, n=20). Cortical field potentials were small and brief, corresponding to the measured glutamate transients. Excitability was then increased by blockade of GABAA receptors. In a disinhibited cortical slice, glutamate transients had a fast onset, but later peak, and were larger in amplitude (, -8.3 ± 0.4% peak Δ FY/FC, 173.1 ± 16.7 ms peak time, 417.3 ± 12.0 ms half-width, p<0.01 compared to control for all measures, n=20). The increase in time resolution gained using line-scans revealed small, fast fluctuations in the fluorescence ratio at the peak and during the recovery of the glutamate transient () suggesting multiple glutamate release events. Next, glutamate transporters were blocked (fast fluctuations occurred in 15/20 stimulus-induced glutamate transients). Blockade of glutamate transporters again decreased the baseline FLII81E - 1μ ratio, indicating gradual extracellular accumulation of glutamate in the absence of functional glutamate reuptake. Under these conditions stimulation caused a larger amplitude (, -11.0 ± 0.4% peak Δ FY/FC, p<0.01, n=20), longer duration (685.3 ± 22.9 ms half-width, p<0.01) decrease in the FLII81E - 1μ fluorescence ratio which peaked later (206.2 ± 11.3 ms, n=20, p<0.05) compared to GABAzine alone. The later peak of glutamate transient was consistent with the result obtained by whole-field scan experiments (). ROI analysis of each cortical layer was then performed on these images to detect layer specific glutamate transients (). Under control conditions, the greatest change in FLII81E - 1μ fluorescence ratio was seen in layers, 4, and 6 (, left, designation of cortical layers is approximate). When GABAA receptors were blocked the amplitude and duration of the glutamate transient was larger and longer in duration in all cortical layers. Blockade of glutamate transporters further prolonged the glutamate transient and increased the maximum glutamate transient amplitude. Cortical field potentials were similar to previously reported results for all pharmacological manipulations tested (). These results show that line scan imaging of FLII81E - 1μ fluorescence allows for higher sampling rates and greatly increases the temporal resolution of this technique.
High-speed (50 Hz) line scan imaging of FLII81E - 1μ glutamate sensor imaging
Calibration of glutamate FRET sensors
Ideally the changes in glutamate FRET sensor fluorescence ratio could be directly converted into changes in glutamate concentration. With this goal in mind a series of glutamate concentrations ranging from 10 μM to 30 mM were applied to slices loaded with FLII81
E - 1μ. Surprisingly, the apparent affinity of the FLII81
E - 1μ for glutamate was approximately 5 mM under these conditions. Very little change in the fluorescence ratio was seen with concentrations less than 1 mM (), while Increasing the applied glutamate concentration from 1 mM to 10 mM caused a drastic decrease in FLII81
E - 1μ fluorescence ratio indicating a threshold effect. These results were at odds with the previously reported affinity of FLII81
E - 1μ for glutamate in free solution (1 μM) (Deuschle, Okumoto, Fehr, Looger, Kozhukh, and Frommer, 2005
). We obtained a standard curve for FLII81
E - 1μ in aCSF in vitro and confirmed that the EC50
was similar to previous reports (, 250 nM). We conclude, therefore, that either the glutamate FRET sensor properties are altered in the extracellular milieu of the brain slice or the glutamate levels within the slice do not reach equilibrium with the applied solution.
Calibration of FLII81E - 1μ glutamate sensor
Two factors might limit our ability to deliver a known concentration of glutamate to the slice: re-uptake of applied glutamate and incomplete penetration of applied glutamate throughout the thickness of the slice. To address the first concern we inhibited glutamate transporters and repeated the glutamate calibration. When glutamate reuptake was pharmacologically inhibited, the apparent affinity of the FRET sensor protein increased to 250 μM and as little as 10 μM was able to induce detectable FRET changes (). Under these conditions, the concentration response curve was fit well by a sigmoidal line and no threshold effect was seen. These results suggest that the capacity for glutamate reuptake is substantial and is capable of buffering applied glutamate concentrations up to 1 mM. The residual difference in apparent affinity of the sensor might result from either incomplete block of EAATS or of other pathways that actively sequester extracellular glutamate.