Imaging ERK and PKA Activation in Single Dendritic Spines during Structural Plasticity

SUMMARY

Extracellular signal-regulated kinase (ERK) and protein kinase A (PKA) play important roles in LTP and spine structural plasticity. While fluorescence resonance energy transfer (FRET)-based sensors for these kinases had previously been developed, they did not provide sufficient sensitivity for imaging small neuronal compartments such as single dendritic spines in brain slices. Here we improved the sensitivity of FRET-based kinase sensors for monitoring kinase activity under two-photon fluorescence lifetime imaging microscopy (2pFLIM). Using these improved sensors, we succeeded in imaging ERK and PKA activation in single dendritic spines during structural long-term potentiation (sLTP) in hippocampal CA1 pyramidal neurons, revealing that the activation of these kinases spreads widely with length constants of more than 10 μm. The strategy for improvement of sensors used here should be applicable for developing highly sensitive biosensors for various protein kinases.

INTRODUCTION

Dendritic spines are considered to be the major loci of synaptic plasticity and function. Typically, a spine is made of a bulbous “head” connected to the dendrite by a thin neck (Arellano et al., 2007, Nagerl et al., 2008, Bourne and Harris, 2008). The neck acts as a diffusion barrier, compartmentalizing biochemical signaling within the spine head (Bloodgood and Sabatini, 2005, Tonnesen et al., 2014). Activity-dependent structural plasticity of dendritic spines plays a central role in synaptic plasticity and, ultimately, learning and memory (Matsuzaki et al., 2004, Hayashi-Takagi et al., 2015, Oh et al., 2013, Hayama et al., 2013). Ca²⁺-triggered signal transduction through numerous signaling proteins, including kinases and small GTPase proteins, is essential for regulating spine structure and synaptic function (Kennedy et al., 2005, Nishiyama and Yasuda, 2015). However, the exact signaling mechanisms that coordinate changes in spine morphology and synaptic strength remain elusive.

Two-photon fluorescence lifetime imaging microscopy (2pFLIM), in combination with FRET biosensors, has been proven to be useful for imaging signaling cascades during structural and functional LTP (Harvey et al., 2008b, Lee et al., 2009, Murakoshi et al., 2011, Zhai et al., 2013, Bosch et al., 2014, Kim et al., 2015, Harward et al., 2016, Hedrick et al., 2016). When single spines are stimulated with glutamate uncaging or electrical stimulation, Ca²⁺ influx through NMDA receptors triggers complex signaling cascades, including the activation of protein kinases and small GTPase proteins, in turn regulating actin polymerization and increasing spine volume and postsynaptic sensitivity to glutamate (Nishiyama and Yasuda, 2015). Among kinases in such signaling cascades, the roles of extracellular signal-regulated kinase (ERK) and cyclic adenosine monophosphate (cAMP)-dependent protein kinase A (PKA) have been known to be critical for many forms of synaptic plasticity and thus extensively studied using pharmacological and biochemical techniques (Kelleher et al., 2004, Thomas and Huganir, 2004, Patterson et al., 2010, Araki et al., 2015, Yagishita et al., 2014, Diering et al., 2014, Li et al., 2013, Hell, 2016). Despite these advances, the spatiotemporal dynamics of ERK and PKA signaling in neurons have been largely unknown, especially in subcellular compartments such as dendritic spines. Recent studies suggest, however, that FRET-based reporters provide a plausible approach to monitoring kinase signaling in living neurons (Harvey et al., 2008a, Zhai et al., 2013, Chen et al., 2014). It has been reported that ERK is activated in the nucleus in response to sLTP of a few dendritic spines, which subsequently turns on transcription factors. However, the sensitivity of the previous kinase sensors was not sufficient to image activities of ERK and PKA in single dendritic spines.

Here we developed highly sensitive sensors for activity of ERK and PKA by optimizing the FRET pair for 2pFLIM. The new FRET pair improved the sensitivity by about 2–3 fold and allowed us to measure ERK and PKA activity during structural plasticity in single dendritic spines and as they spread along the dendrite. Furthermore, to ensure that the spreading of the signal derived from activity of ERK/PKA and not sensor diffusion, we fused sensors to domains of PSD-95, effectively immobilizing the sensors (Sturgill et al., 2009). Using these new tools, we demonstrated that both ERK and PKA showed high degrees of spreading during spine structural plasticity. This may explain their roles in synapse-to-nucleus signaling during synaptic plasticity (Hardingham et al., 2001, Paul et a., 2003, Sindreu et al., 2007, Karpova et al., 2013, Lignitto et al., 2011, Vitolo et al., 2002, Smit et al., 2006).

RESULTS

ERK Sensor Design and Characterization

To develop a highly sensitive ERK activity sensor, we extensively modified a previously published ERK kinase activity reporter (EKAR) (Harvey et al., 2008a) (Fig. 1A, B). EKAR is composed of a fluorescent protein-based FRET pair (mCerulean-mVenus or mRFP1-EGFP), a substrate phosphorylation peptide containing mitogen-activated protein kinase (MAPK) target sequence (PDVPRTPVGK) and docking site (FQFP), and the proline-directed WW phospho-binding domain (Harvey et al., 2008a). It has been demonstrated that the insertion of a flexible EV linker (116–244 amino acids) (Komatsu et al., 2011) between WW domain and substrate phosphorylation peptide improves the sensitivity of EKAR when the Ypet-ECFP pair is used (EKAREV) (Komatsu et al., 2011). Thus, we started from this construct with nuclear export signal (NES) to measure EKAR activity in the cytosolic domain (EKAREV-cyto). To test the sensitivity of EKAR variants, HeLa cells expressing these sensors were stimulated with epidermal growth factor (EGF) (100 ng/ml), which activates ERK signaling through EGF receptors. FRET was quantified and imaged with 2pFLIM (Fig. 1C, D). EGF stimulation triggered an increase in FRET, which is associated with a decrease in fluorescence lifetime of the donor fluorophore (ECFP) (Fig. 1E). First, to optimize the sensor for 2pFLIM, we changed the FRET pair from YPet and ECFP to super Resonance Energy-Accepting Chromoprotein (sREACh) and EGFP (Murakoshi et al., 2008) (EKARsg-cyto; used a different filter). We found that this improved the change in fluorescence lifetime by 1.5 fold. To further increase interaction between EGFP and sREACh, we disabled monomeric mutation (R223F) and introduced additional mutations used for YPet (S208F and V224L) in order to increase interaction between FPs (sREAChet) (Nguyen and Daugherty, 2005). Notably, the sensor with sREAChet (EKARet-cyto) provided ~2 fold more signal in response to EGF stimulation compared to EKARsg-cyto (Fig. 1B, E, F). Overall, EKARet-cyto showed ~3 fold improvement in signal amplitude in cells compared to original EKAREV (Fig. 1B, E, F). As a negative control, we mutated the MAPK phosphorylation site in the substrate (Thr to Ala) (negEKARet-cyto). As expected, this mutant sensor did not show signal in response to EGF (Fig. 1E, F). The basal fluorescence lifetime and the amplitude of response were independent of the concentration of EKARet-cyto in cells (Fig. S1).

Figure 1. Design and Functional Expression of ERK sensors in HeLa cells.

(A) Schematic representation of FRET based ERK sensor, EKARet-cyto. The sensor is composed of phospho-Thr binding domain (WW domain), EV linker, ERK substrate peptide and docking site, sREACh with S208F/R223F/V224L mutations, and EGFP. ERK phosphorylation triggers the association of the WW domain with the substrate peptide, increasing FRET between EGFP and sREAChet. The three-residue mutations in sREACh enhanced FRET signal.

(B) Schematic representation of EKAREV-cyto and EKARet-cyto.

(C, D) HeLa cells expressing EKAREV-cyto (C) or EKARet-cyto (D) were stimulated with 100 ng/ml EGF treatment, followed by inhibition with 50 μM U0126. Cells were imaged at indicated time points.

(E) EGF-induced lifetime change of EKAREV-cyto (n = 12); EKARsg-cyto (n = 4); EKARet-cyto (n = 4) and negEKARet-cyto (n = 18).

(F) Fluorescence lifetime changes in HeLa cells expressing different ERK sensors after EGF administration (12 min). *** P < 0.001, **** P < 0.0001 (ANOVA followed by Tukey’s test). Error bars indicate S.E.M.

See also Figure S1.

Next, we investigated whether EKARet-cyto could be utilized for imaging ERK activity in single dendritic spines. We biolistically transfected neurons with EKARet-cyto in mouse organotypic hippocampal slices. We found that the sensors exhibited uniform expression in dendrites and spines (Fig. 2A). To activate ERK, we applied phorbol-12-myristate-13-acetate (PMA) (4 μM), an agonist of protein kinase C (PKC), a kinase located upstream of ERK (Shalin et al., 2006, Tanaka and Augustine, 2008). EKARet-cyto displayed a robust signal after PMA administration in both spines and dendrites, and the signal was reversed by the highly specific MEK inhibitor 1,4-Diamino-2,3-dicyano-1,4-bis[2-aminophenylthio]butadiene (U0126), suggesting that EKARet-cyto specifically measures ERK activity (Fig. 2A, B). In addition, we developed a variant of EKARet that is accumulated in the nucleus by replacing the NES sequence by the nuclear localization signal (NLS) sequence (EKARet-nuc). After adding PMA, ERK was activated in the nucleus (Fig. 2C). Therefore, EKARet detects ERK activation in neurons with the sensitivity sufficient to examine ERK activity in single dendritic spines.

Figure 2. Spatiotemporal dynamics of ERK activation in dendritic structures.

(A) Fluorescence lifetime images of EKARet-cyto dendritic structures before and after PMA and U0126 application.

(B) Time course of EKARet-cyto activity in spines (average of 5 spines) and their parent dendrite in (A) in response to bath application of PMA and, subsequently, U0126. On average, fluorescence lifetime changed by 0.128 ±.0.020 ns (spines, 4 neurons) and 0.128 ±.0.018 ns (dendrites) in response to PMA.

(C) Fluorescence lifetime images of EKARet-nuc in the nucleus before and after PMA application.

ERK Activation during Spine Structural Plasticity

Next, we studied ERK activity during spine structural plasticity using EKARet-cyto. Two-photon glutamate uncaging at a spine causes calcium influx through NMDA receptors, leading to spine cytoskeleton remodeling, thereby inducing sLTP in the stimulated spines (Matsuzaki et al., 2004, Nishiyama and Yasuda, 2015). To determine the spatiotemporal dynamics of ERK activity during this process, we imaged EKARet-cyto in dendritic segments of CA1 pyramidal neurons with 2pFLIM. We applied a low frequency train of two-photon glutamate uncaging pulses (60 pulses at 1Hz) to a single dendritic spine in the absence of Mg²⁺ (Matsuzaki et al., 2004). ERK was activated in both the stimulated spine and the adjacent dendrite in response to glutamate uncaging (Fig. 3A–E). The activation of ERK reached a plateau within ~5 min and returned back to baseline at around 20 min. In parallel with ERK activation, the spine volume increased after glutamate uncaging by 124 ± 35% (mean ± S.E.M., the standard error of the mean) within ~5 min, followed by a sustained phase of volume increase lasting more than ~30 min (43 ± 12 %; average of volume change between 15–40 min) (Fig. 3F). There was no detectable gradient of ERK activation between the stimulated spines and the nearby dendritic shafts (Fig. 3A, C, D), indicating that ERK activation is not restricted to the stimulated spines. ERK activation further spread along the shaft over at least 12 μm (Fig. 3B). We observed a slight delay in the onset of ERK activity in dendritic segments far from the stimulated spine (13 μm; Fig. 3C) but in general the ERK activity increased at a similar rate in the spine and along the adjacent dendrite (Fig. S2). This is consistent with the fast diffusion of ERK (diffuses ~50 μm in 1 min; (Wiegert et al., 2007)) compared to the onset of activation (~2 min). Inhibiting NMDA receptors with 2-Amino-5-phosphonopentanoic acid (AP5) abolished glutamate uncaging-induced ERK activation and structural plasticity (Fig. 3E–I), suggesting that NMDA receptors are necessary for these events. Changes in fluorescence lifetime of EKARet-cyto increased linearly up to ~30 pulses and then saturated (50% saturation at ~ 30 pulses) (Fig. S3).

Figure 3. ERK activation and effect of AP5 during sLTP of single dendritic spines induced with 2-photon glutamate uncaging.

(A) Representative fluorescence lifetime images of ERK activation during glutamate uncaging. Stimulated spine was marked with the white arrowhead.

(B) ERK activation during sLTP along the stimulated dendrite as a function of the distance from the base of the stimulated spine (n = 12/4; spines/neurons) (stimulated spine, red circle; dendrites, black circle).

(C) Average time course of fluorescence lifetime change of EKARet-cyto in stimulated spines (red line), adjacent dendrites (0 μm, blue line), adjacent spines (green line) and dendritic segments 13 μm away from the stimulated spine (black line) during spine structural plasticity induced with 2-photon glutamate uncaging. Same data set as in B. The inset shows a closer view of the initial 6 min.

(D) Quantification of fluorescence lifetime change in (B) and (C) at indicated time points.

(E) Average time-course of fluorescence lifetime change of EKARet-cyto in the stimulated spines and the adjacent dendrite in the absence (n = 5/3) or presence (n = 11/3) of AP5.

(F) Averaged time course of volume change in the stimulated spines and adjacent dendrites in the absence (n = 5/3) or presence (n = 11/3) of AP5 correlated to (E). The glutamate uncaging was indicated as the black bar in (E) (F).

(G–I) Quantification of peak fluorescence lifetime change of EKARet-cyto (G) correlated to (E), spine volume change during the transient phase (6 min; H) and spine volume change during the sustained phase (average of 27–36 min; I) of sLTP in the stimulated spines and adjacent dendrites in the absence (n = 5/3; spines/neurons) or presence (n = 11/3; spines/neurons) of AP5 correlated to (F). Spine volume changes were measured as changes in fluorescence intensity of EKARet-cyto in the stimulated spines. All data were presented as mean ± S.E.M. (error bars). Statistical significance was tested with one-way ANOVA followed by Tukey’s test or t-test (ns P > 0.05, *** P < 0.001, **** P < 0.0001). Error bars indicate S.E.M.

See also Figures S2, S3

EKARet Tethered to PSD during Spine Structural Plasticity

It is possible that the spatially broad activity of ERK measured by EKARet-cyto could be caused by the diffusion of the sensor itself, rather than the spreading of the ERK activity. To address this ambiguity, we restricted the diffusion of the sensor by fusing the sensor with PDZ-1 and PDZ-2 of PSD-95 (EKARet-PSD) (Fig. 4A). As expected, EKARet-PSD was concentrated in spines (Fig. 4B, Fig. S4A, B). To measure the diffusion of these sensors, we performed fluorescence recovery after photobleaching (FRAP) experiments. When EKARet-PSD was bleached in a single spine with 2-photon laser, the fluorescence in the spine showed extremely slow recovery with more than 50% of bleached fluorescence not recovered in more than ~40 min (Fig. S4C, E, G), consistent with reported recovery rates of PSD-95 in vitro and in vivo (Sturgill et al., 2009, Kuriu et al., 2006, Sharma et al., 2006, Gray et al., 2006). In contrast, we observed immediate recovery of fluorescence of EKARet-cyto (< 1 min; Fig. S4D, F, H). These results suggest that the diffusion of EKARet-PSD is much slower than that of EKARet-cyto. In HeLa cells, EKARet-PSD showed similar basal fluorescence lifetime with EKARet-cyto, but slightly smaller changes in response to stimulation (Fig. S1). Although the majority of EKARet-PSD is localized to PSD puncta, a small fraction of cytosolic component is present. The cytosolic population showed higher basal fluorescence lifetime (lower FRET) and smaller responses to stimulation compared to that in puncta (Fig. S8).

Figure 4. Design and characterization of PSD-PDZ1–2 tethered EKARet-PSD in response to glutamate uncaging.

(A) Schematic representation of EKARet-PSD.

(B) Fluorescence lifetime images of EKARet-PSD expressed in a CA1 pyramidal neuron during sLTP induced with 2-photon glutamate uncaging at a spine (white arrow head). The stimulation causes ERK activation in adjacent spines (white arrows).

(C) Averaged time courses of fluorescence lifetime change in puncta at various distances from the stimulated spines (n = 7/4; puncta/neurons). The inset to (C) shows a closer view of the initial 6 min. Error bars indicate S.E.M.

See also Figure S1, S4 and S8.

Using this immobile sensor, we measured ERK activity during sLTP induced with 2-photon glutamate uncaging. One drawback of the PSD-fused sensor is that, due to differences in basal fluorescence lifetime in PSD and the cytosol, changes in the ratio of the sensor population in the cytosol and that in the PSD likely cause changes in fluorescence lifetime in spines (spines are smaller than the spatial resolution of 2p microscopy). Thus, this sensor is not suitable to quantify ERK signal in spines undergoing structural plasticity. We quantified ERK activity in puncta located at different distances from the base of the stimulated spines (not in stimulated spines). We found that ERK activity measured with EKARet-PSD spreads more than ~15 μm (Fig. 4B, C), showing the spatial spreading similar to that of EKARet-cyto. These results indicate that the spreading of ERK activity measured with EKARet-cyto is not due to the diffusion of the sensor, but rather due to the diffusion of kinase activity itself.

PKA Activation in Single Dendritic Spines

Using the same strategy of designing ERK sensors, we developed a new PKA sensor AKARet-cyto by replacing the FRET pair of previously published AKAR3EV (Komatsu et al., 2011) to the EGFP-sREAChet pair and inserting NES (Fig. 5A). To test the sensitivity of the sensor, we transfected it in HeLa cells and applied an adenylyl cyclase activator forskolin (50 μM) and an inhibitor of cAMP-degrading phosphodiesterase (PDE) 3,7-Dihydro-1-methyl-3-(2-methylpropyl)-1H-purine-2,6-dione (IBMX) (100 μM) (Fig. 5B, C). The new sensor AKARet-cyto showed ~1.6 fold enhanced signal compared to AKAR3EV-cyto in HeLa cells. Its basal fluorescence lifetime and signal increase was independent of its concentrations in cells (Fig. S1D, E). The sensor with a point mutation to the phosphorylation site (negAKARet-cyto) did not respond to the increase of intracellular cAMP (Fig. 5C, S5A), illustrating the specificity of the sensors. The PKA inhibitors — 10 μM H-89 or 4 μM KT-5720 — reversed the change in fluorescence lifetime induced by forskolin, suggesting that the sensor is selective and reversible (Fig. S5B, C). We further examined the sensitivity of AKARet-cyto by transfecting the sensor in CA1 pyramidal neurons in organotypic hippocampal slices and applying forskolin and IBMX to activate PKA. We found that this treatment induced a robust decrease in fluorescence lifetime (increase in FRET) in spines and dendrites (Fig. 5D, E), indicating that the sensitivity of the sensor is sufficiently high to detect activity of PKA in single dendritic spines.

Figure 5. Design and characterization of PKA sensors in HeLa cells and neurons.

(A) Schematic representation of FRET-pair based PKA sensors: AKAR3EV-cyto, AKARet-cyto.

(B) Fluorescence lifetime images of AKAR3EV-cyto and AKARet-cyto in HeLa cells before and after forskolin and IBMX administration.

(C) Averaged time courses of fluorescence lifetime change of AKAR3EV-cyto (n = 6), AKARet-cyto (n = 4) and negAKARet-cyto (n = 4) in response to forskolin and IBMX application.

(D) Fluorescence lifetime images of AKARet-cyto before and after forskolin and IBMX administration in neurons.

(E) Averaged time courses of fluorescence lifetime change of AKARet-cyto in spines (n = 28/4; spines/neurons) and their parent dendrites (n = 4/4; dendrites/neurons). Data are presented as mean ± S.E.M. (error bars).

See also Figure S1 and S5.

Using the new PKA sensor, we imaged the spatiotemporal pattern of PKA activity in response to glutamate uncaging (60 pulses at 1 Hz) applied near single dendritic spines. Glutamate uncaging induced a transient PKA activation before the signal returned to baseline within ~5 min (Fig. 6A–C). We observed that PKA activation spread more than 10 μm (Fig. 6A–D; Fig. S6). Compared to ERK activation, PKA activation lasted for a shorter time and showed a sharper spatial gradient along the dendrite (Fig. 3, 6, S6). Changes in fluorescence lifetime of AKARet-cyto increased linearly to the number of uncaging pulses up to ~30 pulses and saturated (50% saturation at ~ 30 pulses) (Fig. S7). AP5 abolished glutamate uncaging-induced PKA activation as well as structural plasticity (Fig. 6E–I), suggesting that NMDA receptor is required for PKA activation.

Figure 6. Spatiotemporal PKA activation after glutamate uncaging at a single spine.

(A) Fluorescence lifetime images of AKARet-cyto in a secondary dendrite of a CA1 pyramidal neuron during sLTP.

(B) PKA activation along the dendrite at various distances from the stimulated spine (n = 11/3) (stimulated spine, red circle; dendrites, black circle).

(C) Average time course of fluorescence lifetime change of AKARet-cyto in stimulated spine (red line), adjacent dendrite (0 μm, blue line), adjacent spine (green line) and dendritic segments 13 μm away from the stimulated spine (black line) during spine structural plasticity induced with 2-photon glutamate uncaging (n = 11/3), correspond to (A) and (B). The inset to (C) shows a closer view of the initial 6 min.

(D) Quantification of fluorescence lifetime change in (C) at indicated time points.

(E) Averaged time courses of fluorescence lifetime changes of AKARet-cyto in the stimulated single spines, adjacent dendrites during sLTP induced in a single spine (n = 7/3). AP5 eliminated PKA activation (n = 11/3).

(F) Volume change of the stimulated spines and adjacent dendrites measured as changes in fluorescence intensity of AKARet-cyto (n = 7/3; spines/neurons), AP5 eliminated spine volume change (n = 11/3; spines/neurons), correlated to (C).

(G–I) Quantification of peak fluorescence lifetime change of AKARet-cyto (G) correlated to (E), spine volume change during the transient phase (4 min; H) and spine volume change during the sustained phase (average of 27–36 min; I) of sLTP in the stimulated spines and adjacent dendrites in the absence (n = 7/3; spines/neurons) or presence (n = 11/3; spines/neurons) of AP5 correlated to (F). All data were presented as mean ± S.E.M. (error bars). Statistical significance was tested with one-way ANOVA followed by Tukey’s test or t-test (ns P > 0.05, * P < 0.05, *** P < 0.001, **** P < 0.0001).

See also Figure S6 and S7.

Using the same strategy of designing EKARet-PSD, we also developed a low mobility PKA sensor by fusing PDZ1–2 domain of PSD-95 to the N-terminal of AKARet (AKARet-PSD; Fig. 7A). Similar to the cytosol sensors, the basal lifetime of both AKARet-PSD was independent of the sensor concentration (Fig. S1D). However, the basal fluorescence lifetime of AKARet-PSD was higher and its response was smaller compared to that of AKARet-cyto (Fig. S1E, Fig. S8). As expected, AKARet-PSD was mainly localized in spines (Fig. 7B, S9A). As with EKARet-PSD, the lifetime of AKARet-PSD in puncta was lower (higher FRET) than the cytosol in neurons (Fig. 7B) and HeLa cells (Fig. S8B, D). As with EKARet-PSD, the diffusion of AKARet-PSD was much slower than that of AKARet-cyto (Fig. S9C, D). Using this immobile sensor, we measured PKA activity in puncta during spine structural plasticity (Fig. 7C). We found that the signals spread similarly to that of AKARet-cyto, suggesting that the spreading of AKARet signal is not due to the diffusion of the sensor, but rather due to the spreading of PKA activation.

Figure 7. Design and characterization of PSD-PDZ1–2 tethered AKARet-PSD during glutamate uncaging.

(A) Schematic representation of AKARet-PSD.

(B) Fluorescence lifetime images of AKARet-PSD during sLTP in a secondary dendrite of a CA1 pyramidal neuron induced in a single dendritic spine with 2-photon glutamate uncaging (white arrowhead). The stimulation causes PKA activation in adjacent spines (white arrows).

(C) Averaged time courses of fluorescence lifetime changes in puncta at various distances from the stimulated spines (n = 9/4; puncta/neurons). Data were presented as mean ± S.E.M. (error bars).

See also Figure S1, S8, and S9

DISCUSSION

In this study, we improved the sensitivity of FRET-FLIM-based sensors for ERK and PKA by using sREAChet, a modified dark acceptor, and EGFP as a new FRET pair. The new FRET pair increased the fluorescence lifetime change of the sensors by approximately 2–3 fold. The new sensors allowed us to visualize spatiotemporal dynamics of ERK and PKA in single dendritic spines undergoing structural plasticity in organotypic hippocampal slices. In particular, we have demonstrated that activity of ERK and PKA spreads out of the stimulated spines and invades neighboring spines and dendritic shafts over more than ~13 μm. Extrapolating this data, we predict that the spreading likely extends more than several tens of micrometers, potentially covering an entire secondary dendritic branch. The strong spreading may play important roles in synapse-to-nucleus signaling (Zhai et al., 2013) as well as heterosynaptic plasticity (Harvey et al., 2008b; Nishiyama and Yasuda, 2015) and other signals occurs in dendritic shafts such as exocytosis of glutamate receptors and synthesis of new proteins (Makino et al., 2009; Patterson et al., 2010; Govindarajan et al, 2011).

With a diffusible substrate-based sensor, it is difficult to distinguish whether spreading activity is caused by the diffusion of activated sensors or the diffusion of the kinase of interest. In addition, the spatiotemporal dynamics of signaling activity may be significantly perturbed by the presence of the sensor, since the sensor interacts with endogenous signaling proteins (Augustine and Neher, 1992). Thus, the interpretation of signals often requires modeling of the biochemical process of the sensors (Yasuda and Murakoshi, 2011, Lee and Yasuda, 2009). Diffusible and immobile sensors, in general, may cause signaling that spreads more or less, respectively (Augustine and Neher, 1992). By creating sensors with different mobility, we can address these concerns. In addition to cytosolic sensors, we developed sensors tethered to PSD-95 by fusing the sensor with PDZ1–2 of PSD-95, which largely suppressed the mobility of the sensors. The activity of ERK and PKA measured by these immobile sensors also displayed spreading similar to that of the cytosolic sensors, suggesting that spreading of the activity of endogenous ERK and PKA occurs under these conditions.

In summary, our novel ERK and PKA sensors provide useful tools for imaging ERK and PKA activity in small neuronal compartments such as dendritic spines. Moreover, the same donor-acceptor pair should be applicable to improve the sensitivity of sensors for a broad spectrum of kinases.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for reagents may be directed to, and will be fulfilled by the lead contact Ryohei Yasuda (ryohei.yasuda@mpfi.org).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

C57BL/6 mice (timed-pregnant females at gestational day 16) were obtained from Charles River and pups were used for the preparation of organotypic hippocampal slice culture at postnatal days 4–6. All animal experimental procedures were approved by the Max Planck Florida Institute for Neuroscience Institutional Animal Care and Use Committee (IACUC) and were performed in accordance with national guidelines (National Institutes of Health). HeLa cells were purchased from ATCC.

Cell Culture

HeLa cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (Life Technologies) containing 10% fetal bovine serum (FBS) at 37 °C in a 5% CO₂ atmosphere. DNA of each sensor was transfected into cells using Lipofectamine 2000 (Invitrogen) (3 μg DNA per 35 mm dish). Cells were washed 3 times using phosphate-buffered saline (PBS) (ThermoFisher Scientific) containing 1.06 mM KH₂PO₄, 155.17 mM NaCl, 2.97 mM Na₂HPO₄·7H₂O 8–12 hrs post transfection. The culture medium was replaced to 0.2% FBS DMEM for overnight incubation. Before imaging, cells were washed by mammalian cell imaging buffer (114 mM NaCl, 2.2 mM KCl, 22 mM NaHCO₃, 1.1 mM NaH₂PO₄, 2 mM glucose, 25 mM HEPES, pH 7.2~7.3, with 2 mM CaCl₂ and 2 mM MgCl₂) for three times and incubated for 30 min at room temperature. To quantify the concentration of sensors in HeLa cells, we measured the fluorescence intensity of EKARet-cyto, EKARet-PSD, AKARet-cyto and AKARet-PSD, in direct proportion to the measured fluorescence intensity of 10 μM or 2 μM purified EGFP under two-photon microscope.

Organotypic Hippocampal Slice Cultures

Hippocampi from postnatal day (P) 4–6 mice were dissected and cultured in culture medium composed of 8.38 g/L MEM, 30 mM HEPES, 5.2 mM NaHCO₃, 12.9 mM D-Glucose, 4.25 mM Ascorbic Acid, 1 mM L-Glutamine, 1 μg/ml insulin, 1 mM CaCl₂, 2 mM MgSO₄, 20% horse serum. Gold microcarrier particle (1.0 μm, Bio-Rad) was coated with plasmid DNA encoding each sensor (DNA/gold = 8.33 μg/mg). Biolistic transfection was performed using a Helios gene gun (Bio-Rad) at 180 psi helium pressure at days in vitro (DIV) 9–11 for ERK and 8–10 for PKA. Imaging was performed 2–4 days after the transfection. Before imaging, the slices on the insert were incubated in HEPES-buffered Artificial Cerebrospinal Fluid (HACSF) buffer (130 mM NaCl, 2.5 mM KCl, 2 mM NaHCO₃, 1.25 mM NaH₂PO₄, 25 mM D-glucose, 20 mM HEPES, with 4 mM CaCl₂ and 1 μM tetrotoxin (TTX), pH 7.3, osmolarity 300–310, pre-aerated with 95% O₂ and 5% CO₂ for at least 10 min) for ~1 hr at room temperature. During imaging, the slices were immersed in ACSF (127 mM NaCl, 2.5 mM KCl, 25 mM NaHCO₃, 1.25 mM NaH₂PO₄, 25 mM D-glucose, with 4 mM CaCl₂, 1 μM TTX, osmolarity 300~310, aerated with 95% O₂ and 5% CO₂). For glutamate uncaging experiments, 4 mM 4-Methoxy-7-nitroindolinyl-caged-L-glutamate (MNI-caged-L-glutamate; Tocris) was added to ACSF.

METHOD DETAILS

Plasmids

The constructs encoding EKAR (mRFP1-EGFP) (Harvey et al., 2008a), sREACh (Murakoshi et al., 2008), PSD-95-EGFP were generated previously. The constructs encoding EKAREV (YPet-ECFP) and AKAR3EV (YPet-ECFP) were generous gifts from Dr. Michiyuki Matsuda (Kyoto University, Japan) (Komatsu et al., 2011).

To create EKARsg, ECFP and YPet in EKAREV was replaced with EGFP and sREACh, respectively. sREAChet (sREACh[S208F, R223F, V224L]) was generated by site-directed mutagenesis to sREACh. sREACh of EKARsg was replaced by sREAChet to produce EKARet. NES was added to the C-tail of EKARet to generate EKARet-cyto. AKARet-cyto was constructed using the same strategy.

To generate EKARet-PSD, PSD-95 was inserted before the N-terminal of EKARet with a 24-amino acid linker (LGILQSTVPRARDPPVATMACQAS) between PSD-95 and sREAChet, and then the PDZ3, SRC Homology 3 (SH3) and guanylate kinase (GK) domains were deleted. The C-tail (IYHKVKRVIEDLSGPYIWVPARER) of PSD-95 remained in the construct. The same strategy was used to generate AKARet-PSD.

Imaging

Two-photon fluorescence lifetime imaging was performed as described previously (Yasuda et al., 2006, Zhai et al., 2013). We used a custom-built two-photon microscope (based on Olympus) equipped with two Ti: sapphire lasers (Coherent), of which one was turned to 920 nm to excite the donor of FRET sensor, and the other was turned to 720 nm to uncage glutamate. The intensity of the lasers was controlled by Pockels cells (Conoptics). EKARsg-cyto and EKARet were imaged with photo-multiplier tubes (Hamamatsu) placed after wavelength filters (ET520/60-2p and ET620/60-2p, Chroma) and a dichroic mirror (Q565). For EKAREV-cyto (YPet-ECFP), different filters and dichroic mirror were used (ET480/40 and ET535/50, Q505). Fluorescence emission was collected with a 60× objective lens (1.0 NA, Olympus). The fluorescence lifetime images were acquired with time-correlated single photon counting method (Becker and Hickl). We analyzed fluorescence lifetime images as described previously (Yasuda et al., 2006, Harvey et al., 2008a, Zhai et al., 2013). Briefly, the mean fluorescence lifetime τ was calculated using the equation:

$$\mathrm{\tau }=\int t·F(t)\mathrm{d}t/\int F(t)\mathrm{d}t-t_0$$

where <t> is the mean photon arrival time, t₀ is the offset arrival time, and F(t) is the fluorescence lifetime decay.

Two-photon Glutamate Uncaging

The two-photon Ti:sapphire laser turned at 720 nm was used for glutamate uncaging. Spine structural plasticity was induced by applying pulses of the focused laser near a spine of interest (6 ms, 60 pulses at 1 Hz, 3.6–3.9 mW). During the experiments, the temperature of ACSF was kept at 26 °C using Dual Channel Heater Controller (Warner Instruments). Changes in spine volume were quantified as changes in the fluorescence intensity of the sensors expressed within the stimulated spine.

To measure the correlation between the number of uncaging pulses and changes in fluorescence lifetime of EKARet-cyto and AKARet-cyto, we applied 10, 30 or 60 uncaging pulses to different spines in a neuron. The relationship between changes in fluorescence lifetime and the number of uncaging pulses were fitted with a linear saturation equation (similar to the Michaelis-Menten kinetics equation):

$$\mathrm{\Delta }\mathit{Lifetime}=\frac{\mathrm{\Delta }{\mathit{Lifetime}}_{\mathit{max}}}{1+\frac{{EC}_{50}}{\mathit{Uncaging}\mathit{pulses}}},$$

where ΔLifetime_max is the saturated change in fluorescence lifetime and EC50 is the number of uncaging pulses that causes 50% saturation. We obtained ΔLifetime_max = 0.25 and EC₅₀ = 29 pulses for for EKARet-cyto and ΔLifetime_max = 0.13 and EC₅₀ = 28 pulses for AKARet-cyto.

FRAP

The sensor mobility was obtained by imaging green fluorescence intensity in several spines. To photo-bleach fluorescence in a spine, the spine was scanned with the imaging laser (920 nm, 5–7 mW) for 10 s.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics

All the data were presented as mean ± S.E.M., and asterisks indicated the statistical significance (ns P > 0.05, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001). Statistical analysis was performed using t-test for comparison between two groups and one-way ANOVA followed by Tukey’s test for comparison between more than two groups (Prism 6, GraphPad). n denotes the number of cells for the experiments using HeLa cells and the numbers of spines/cells or dendrites/cells for experiments using neurons.