Rac1 is a downstream effector of PKCα in structural synaptic plasticity

Structural and functional plasticity of dendritic spines is the basis of animal learning. The rapid remodeling of actin cytoskeleton is associated with spine enlargement and shrinkage, which are essential for structural plasticity. The calcium-dependent protein kinase C isoform, PKCα, has been suggested to be critical for this actin-dependent plasticity. However, mechanisms linking PKCα and structural plasticity of spines are unknown. Here, we examine the spatiotemporal activation of actin regulators, including small GTPases Rac1, Cdc42 and Ras, in the presence or absence of PKCα during single-spine structural plasticity. Removal of PKCα expression in the postsynapse attenuated Rac1 activation during structural plasticity without affecting Ras or Cdc42 activity. Moreover, disruption of a PDZ binding domain within PKCα led to impaired Rac1 activation and deficits in structural spine remodeling. These results demonstrate that PKCα positively regulates the activation of Rac1 during structural plasticity.

In order to study signaling in single spines during structural plasticity, we combined two-photon release of caged glutamate 25 , fluorescence resonance energy transfer (FRET)-based sensors, and two-photon fluorescence lifetime imaging microscopy (2pFLIM) 26 to monitor the dynamics of intracellular signaling events with high spatiotemporal resolution. Specifically, using previously published FRET sensors, we monitored the spatiotemporal activation of the actin-regulating small GTPases, including Ras [27][28][29] , Cdc42 15 and Rac1 19 during the induction of structural long-term potentiation (sLTP). Briefly, these sensors are composed of two components: (1) full length GTPase fused to green fluorescent protein (GTPase-eGFP), and (2) a specific GTPase binding domain of a downstream effector fused to two copies of red fluorescent protein (mRFP-effector-mRFP) (Fig. 1a). Activation of the GTPase increases the affinity between the two components of the sensor and the FRET between the fluorophores. Using 2pFLIM we can measure small GTPase activation by monitoring changes in the binding of the sensor components (binding fraction) through quantitatively measuring decreases in the fluorescence lifetime of GFP.
We transfected organotypic hippocampal slices from wildtype (WT) or PKCα knockout (KO) mice with small GTPase sensors and imaged CA1 pyramidal neurons using 2pFLIM. In response to glutamate uncaging targeted to a single dendritic spine (30 pulses at 0.5 Hz), the stimulated spine from WT slices rapidly enlarged by ~150% (transient phase) and persisted with an increased volume of ~40% (sustained phase) lasting at least 25 min (Fig. 1b,c). This sustained increase in the spine volume is highly correlated with increases in the functional strength of the stimulated spine 28,30 . Consistent with previous work, hippocampal CA1 neurons from PKCα KO mice showed a deficit in sLTP 14 . Moreover, sLTP of stimulated spines expressing Rac1 sensor was consistent with spines expressing eGFP alone (Fig. 1c), demonstrating, and consistent with previous characterization 15,19 , that effects of sensor overexpression on endogenous signaling are minimal. When sLTP was induced in single dendritic spines, we observed rapid and sustained activation of Rac1 (Fig. 1b,d) in the WT slices that was consistent with previously reported findings 19 . However, we found that in the absence of PKCα, Rac1 activity is lower at baseline and its activation is significantly attenuated during sLTP (Fig. 1d,e). Rac1 activity in WT mice was restricted to the stimulated spines at early time points (2-4 min), but spread into the dendrite and nearby spines at later time points (10 -20 min), consistent with a previous study 19 . In PKCα KO mice, Rac1 activity in both stimulated spines and adjacent dendrites were attenuated (Fig. 1e,f). Thus, PKCα positively regulates Rac1 activation to facilitate plasticity.
In order to test the specificity of PKCα signaling toward Rac1, we tested whether the plasticity-induced activation of other small GTPases involved in plasticity, Ras or Cdc42 1,31,32 , was also impaired in the absence of PKCα. Impairment of the functionally relevant sustained phase of sLTP was observed in PKCα KO neurons transfected with Ras or Cdc42 sensor (Fig. 2a,c). However, the activation of these two small GTPases during sLTP were not affected by loss of PKCα (Fig. 2b,d). This demonstrates that Rac1 but not Ras or Cdc42 is downstream of PKCα during sLTP.
We next investigated whether the PDZ binding domain of PKCα, which was shown to be critical for its isozyme-specific signaling in spine plasticity 14 , was required for downstream regulation of Rac1 (Fig. 3a). PKCα or PKCα lacking its PDZ binding domain was sparsely and postsynaptically expressed alongside the Rac1 sensor in hippocampal slices from PKCα KO mice. We found impairment of sLTP (Fig. 3b) and Rac1 activation (Fig. 3c) only when the PDZ binding domain was disrupted. This finding suggests a crucial role for PDZ binding domain of PKCα in Rac1 activation during sLTP.

Discussion
In this study, we have described that PKCα regulates the activation of Rac1, but not Ras or Cdc42, during sLTP of dendritic spines (Fig. 3d). This modulation relies on PKCα's PDZ-binding motif, which may localize PKCα to a signaling complex scaffold that also recruits Rac1 GTPase-activating proteins (GAPs) or Guanine nucleotide-exchange factors (GEFs). Although the specific mechanism through which PKCα regulates Rac1 activity remains to be determined, potential scaffolds include PICK1, SAP-97 or PSD-95, which can interact with PKCα through its C-terminal PDZ binding domain and have been implicated in plasticity of spines [33][34][35][36][37][38] .
The regulation of Rac1 during plasticity through GAPs and GEFs is highly complex in order to allow for tightly-regulated remodeling of various cytoskeletal domains in a precise spatial and temporal pattern 39 . Rac1 has been shown to be localized to signaling complexes that scaffold upstream activators, small GTPases and downstream effectors, enabling subcompartmentalization of functional output 40 . Rac1 activation promotes F-actin polymerization that facilitates spine growth and stabilization through the activation of major downstream targets p21-activated kinase (PAK), LIM kinase I (LIMKI) and WAVE regulatory complex (WRC) 31,41 . The phosphorylation of PAK and LIMKI leads to inactivation of actin depolymerizing factor ADF/cofilin, which in turn stimulates F-actin polymerization 1,31 . Additionally, Rac1 regulates actin branching through activation of WRC which drives Arp2/3 41,42 . Interestingly, PICK1 has also been shown to bind and regulate Arp2/3 43 .
While PKCα could modulate Rac1 through inhibition of a GAP or Rho GDP dissociation inhibitors (RhoGDIs) [44][45][46] , much of the specificity of Rac1 GTPase signaling is reportedly regulated by GEFs [46][47][48] . At least ten different RhoGEFs have been identified to be localized to the post-synaptic density 49 . Of these, Kalirin 7, Tiam1 and B-Pix all show preferential GEF activity for Rac1, are able to be phosphorylated by classic PKC isozymes in-vitro and are required for sLTP of spines [50][51][52][53][54][55][56][57] . Tiam1 contains a PDZ domain that could potentially interact with PKCα, however the structure of its PDZ domain does not predict PKCα as a preferred binding partner 58 . Kalirin7 and B-Pix also each contain a PDZ binding domain, which could localize them to PDZ containing-scaffolds together with activated PKCα 57,59,60 . Indeed Kalirin-7 binds scaffolds including PSD-95, PICK1 and SAP97 in vitro. Only its interaction with PSD-95 has been confirmed in hippocampal neurons 61 .
It is important to note that these experiments rely on overexpression of small GTPase sensors. Importantly the effects of sensor overexpression on volume change of stimulated spines, sensor activation and kinetics have been well characterized and show minimal effects in WT neurons 15,19,28 . However, as observed in Fig Cdc42 may partially rescue the impairment of plasticity in a KO background. This finding is interesting given the known crosstalk between Cdc42 and Rac1 in regulating actin dynamics 62 . Rac1 and Cdc42 share several common effectors 63 that are relevant to structural plasticity of spines. For example, both activated Rac and Cdc42 can increase PAK1 association with LIMK and LIMK-dependent phosphorylation of cofilin 64 . Cofilin phosphorylation has been shown to be an important regulator of actin remodeling during structural plasticity of spines and is transiently increased during the early phase (~2 min) of plasticity 65  www.nature.com/scientificreports www.nature.com/scientificreports/ some of the PKCα -Rac1 mediated KO deficits. Further study on potential crosstalk between Cdc42 and Rac1, particularly related to cofilin regulation during structural plasticity of spines will be informative. Both Rac1 and PKCα activation are downstream of TrkB receptor activation in the stimulated spine 14,19 (Fig. 3d). One important future direction is to determine if TrkB-dependent activation of Rac1 is solely through PKCα or whether a more complex feedback loop is in play. The increased activity and spreading of Rac1 promoted by PKCα is consistent with an essential role for PKCα in integrating neurotrophic signals to facilitate plasticity. Moreover, this link between PKCα and Rac1 supports a growing understanding of the molecular mechanisms underlying heterosynaptic facilitation, whereby the induction of plasticity in one spine lowers the threshold of plasticity induction in nearby spines 14,28,66 .
This study identified a novel molecular pathway that links short-lived calcium influx and activation of PKCα, to long-lasting Rac1 activation and changes in spine structure. This work, taken together with previous findings that another Ca 2+ -dependent kinase, CaMKII, also activates Rac1 during sLTP 57,66 , Rac1 appears to be a key convergence point of multiple upstream calcium-dependent pathways (Fig. 3d). We anticipate that the better understanding of Rac1 signaling pathway may ultimately help us to uncover the signaling network with which memories are encoded.

Materials and Methods
Animals. All experimental procedures were approved and carried out in accordance to the regulations of the Max Planck Florida Institute for Neuroscience Animal Care and Use Committee in accordance with guidelines by the US National Institutes of Health. P4-P8 mouse pups from both sexes were used for organotypic slices for imaging studies. The genotype of each animal was verified before preparing slices. PKCα KO 129/sv animals were received from Dr. Michael Leitges. Animals were crossed to C57Bl/6N Crl and are on a mixed background. For all the experiments, WT littermates were used as controls for KO animals.
Organotypic hippocampal slice cultures and transfection. Organotypic hippocampal slices were prepared from wildtype or transgenic postnatal 4-8 day old mouse pups of both sexes as previously described 68 . In brief, the animals were anaesthetized with isoflurane, after which the animal was quickly decapitated and the brain removed. The hippocampi were dissected and cut into 350 µm thick coronal hippocampal slices using a McIlwain tissue chopper (Ted Pella, Inc) and plated on hydrophilic PTFE membranes (Millicell, Millipore) fed by culture medium containing MEM medium (Life Technologies), 20% horse serum, 1mM L-Glutamine, 1 mM CaCl 2 , 2 mM MgSO 4 , 12.9 mM D-Glucose, 5.2 mM NaHCO 3 , 30 mM Hepes, 0.075% Ascorbic Acid, 1 µg/ml Insulin. Slices were incubated at 37 °C in 5% CO 2 . After 7-12 days in culture, CA1 pyramidal neurons were transfected with biolistic gene transfer 69 using 1.0 µm gold beads (8-12 mg) coated with plasmids containing 50 μg of total cDNA of interest in the following ratios. Rac1 sensor, donor: acceptor = 1:2; Rac1 sensor plus PKCα, donor: acceptor: PKCα = 1:2:1; Rac1 sensor plus PKCα without PDZ binding domain, donor: acceptor: PKCα-no PDZ domain = 1:2:1; Ras sensor, donor: acceptor = 1:3; Cdc42 sensor, donor: acceptor = 1:1). Neurons expressing all plasmid combinations were imaged 2-5 days after transfection. 2pFLIM. FLIM imaging using a custom-built two-photon fluorescence lifetime imaging microscope was performed as previously described 15 . 2pFLIM imaging was performed using a Ti-sapphire laser (Coherent, Cameleon) at a wavelength of 920 nm with a power of 1.4-1.6 mW. Fluorescence emission was collected using an immersion objective (60×, numerical aperture 0.9, Olympus), divided with a dichroic mirror (565 nm) and detected with two separated photoelectron multiplier tubes placed after wavelength filters (Chroma, 510/70-2p for green and 620/90-2p for red). Both red and green channels were fit with photoelectron multiplier tubes (PMT) having a low transfer time spread (H7422-40p; Hamamatsu) to allow for fluorescence lifetime imaging. Photon counting for fluorescence lifetime imaging was performed using a time-correlated single photon counting board (SPC-150; Becker and Hickl) and fluorescence images were acquired with PCI-6110 (National instrument) using modified ScanImage 70 (https://github.com/ryoheiyasuda/FLIMimage_Matlab_ScanImage). Intensity images for analysis of sLTP volume changes were collected by 128 × 128 pixels as a z stack of three slices with 1 µm separation and averaging 6 frames/slice. Spine volume was measured as the integrated fluorescent intensity of EGFP after subtracting background (F). Spine volume change was calculated by F/F 0 , in which F 0 is the average spine intensity before stimulation.
Two-photon glutamate uncaging. A second Ti-sapphire laser tuned at a wavelength of 720 nm was used to uncage 4-methoxy-7-nitroindolinyl-caged-l-glutamate (MNI-caged glutamate) in extracellular solution with a train of 4-8 ms, 2.8-3.0 mW pulses (30 times at 0.5 Hz) in a small region ~0.5 µm from the spine of interest as previously described 14  www.nature.com/scientificreports www.nature.com/scientificreports/ 2pFLIM analysis. To measure the fraction of donor that was undergoing FRET with acceptor (Binding Fraction), we fit a fluorescence lifetime curve summing all pixels over a whole image with a double exponential function convolved with the Gaussian pulse response function: where τ AD is the fluorescence lifetime of donor bound with acceptor, P D and P AD are the fraction of free donor and donor undergoing FRET with acceptor, respectively, and H(t) is a fluorescence lifetime curve with a single exponential function convolved with the Gaussian pulse response function: in which τ D is the fluorescence lifetime of the free donor, τ G is the width of the Gaussian pulse response function, F 0 is the peak fluorescence before convolution and t 0 is the time offset, and erfc is the complementary error function.
We fixed τ D to the fluorescence lifetime obtained from free eGFP (2.6 ns), and then fixed τ AD to fluorescence lifetime of the donor bound with acceptor (1.1 ns). For experimental data, we fixed τ D and τ AD to these values to obtain stable fitting.
To generate the fluorescence lifetime image, we calculated the mean photon arrival time, <t>, in each pixel as: Then, the mean photon arrival time is related to the mean fluorescence lifetime, <τ>, by an offset arrival time, t o , which is obtained by fitting the whole image: For small regions-of-interest (ROIs) in an image (spines or dendrites), we calculated the binding fraction (P AD ) as: Data with lifetime fluctuations in the baseline that were greater than 0.15 ns were excluded before further analysis. To analyse the spatial spreading of Rac1 activation, contiguous 1 μm diameter ROIs along dendrite from the base of the stimulated spines were analysed for Rac1 activation over time.
Statistical analysis. All values are presented as mean ± SEM unless otherwise noted. Number of independent measurements (n[neurons/spines]) is indicated in figure legends. Unpaired two-tailed student's t test was used for comparing two independent samples. Two-way ANOVA followed by multiple comparison test was used to compare grouped data sets (Prism 6, GraphPad). Data were excluded if obvious signs of poor cellular health (for example, dendritic blebbing, spine collapse) were apparent. In addition, outlier analysis was performed on spine volume data with ROUT (Q = 1%).This analysis led to the exclusion of 2 out of 113 spines from analysis.