A-Kinase Anchoring Protein 79/150 recruits Protein Kinase C to phosphorylate Roundabout receptors

Results: approaches confirm that the AKAP79/150 signaling complex interfaces with the cytoplasmic tail of Roundabout (Robo) receptors. Conclusion: associated protein kinase C facilitates the phosphorylation of the Significance: Kinase anchoring is a mechanism to control the phosphorylation of Robo3.1 within macromolecular assemblies. ABSTRACT Anchoring proteins direct protein kinases and phosphoprotein phosphatases toward selected substrates to control the efficacy, context, and duration of neuronal phosphorylation events. The A-Kinase Anchoring Protein AKAP79/150 interacts with protein kinase A (PKA), protein kinase C (PKC), and protein phosphatase


INTRODUCTION
Axonal guidance cues influence neuronal development by modulating local cell signaling events at growth cones of migrating neurons and at the postsynaptic densities of mature neurons (1)(2)(3). A common feature is the dissemination of extracellular signals through transmembrane receptor-associated complexes that include protein kinases and phosphatases (4)(5)(6). These anchored enzymes often reside on the inner face of the plasma membrane where they respond to the generation of chemical second messengers such as cyclic nucleotides, calcium, and phospholipids (7,8). Scaffolding and anchoring proteins enhance the fidelity of this process by sequestering second messenger responsive kinases and phosphatases within range of particular substrates (9). In some cases, the receptors are the substrates themselves.
For example, A-kinase anchoring proteins (AKAPs) 3 , a burgeoning family of intracellular proteins that sequester protein kinase A, have been shown to modulate the phosphorylation status and activity of several neuronal transmembrane receptors (10,11). Subsequent work has shown that AKAPs coordinate higher order macromolecular complexes of PKA and other signaling molecules at defined subcellular locations (12,13). Consequently, AKAPs augment the specificity of local cell signaling events by scaffolding protein kinases, phosphatases, phosphodiesterases, and small GTPases in proximity to upstream activators and within range of their downstream targets (14)(15)(16)(17). Products of the AKAP5 gene encode a multivalent anchoring protein known as AKAP79/150 (note: AKAP79 is the human form and AKAP150 is the murine ortholog). AKAP79/150 organizes signaling enzymes that participate in the regulation of synaptic ionotropic and metabotropic glutamate receptors as well as other neuronal G-protein coupled receptors (18)(19)(20)(21). A principal molecular role of AKAP79/150 is to tether PKA, PKC, and PP2B within range of these receptors to control changes in synaptic strength. These local signaling events underlie aspects of hippocampal learning and memory, and a myriad of other cellular functions (22,23).
In this report, we show AKAP79/150 binds the Roundabout axon guidance receptors Robo2 and Robo3. AKAP150 co-distributes with both of these Robo receptors in the hippocampus and olfactory tubercle of adult mice. Biochemical and cell-based experiments confirm the AKAP-Robo subcomplex provides a platform for the assembly of larger signaling complexes that include kinases and phosphatases. Accordingly, we demonstrate that AKAP79/150-associated PKC regulates the phosphorylation status of Robo3 in vitro and inside cells.

Mass Spectrometry
Bands of interest from silver stained gels were excised and analyzed using MALDI-TOF by the OHSU proteomics shared resource.
Lysates were centrifuged at 15,000 × g for 20 minutes. Supernatants were incubated on nutator overnight at 4 degrees with either FLAG-agarose (Sigma-Aldrich) or anti-V5 antibody (Invitrogen) and Protein A/G-agarose (Millipore). Beads were washed 2X with HSE buffer supplemented with 10% glycerol and 500 mM NaCl and another 2X with regular HSE buffer. LDS buffer (Invitrogen) was added and samples were run on NuPage gels (Invitrogen).

HEK293 Staining
HEK293 cells were transfected with V5-Robo2 and FLAG-AKAP79 constructs. After 24 hours, coverslips were washed 3X in PBS and fixed for 20 minutes in 4% paraformaldehyde (in PBS) at room temperature. Cells were permeabilized in PBS containing 0.1% Triton X-100 and blocked for 1 hour in PBS supplemented with 10% donkey serum and 0.1% fish gelatin (Sigma). Cells were incubated with antibodies against V5 (Invitrogen; 1:1000) and FLAG (Sigma; 1:1000) overnight at 4 degrees. Coverslips were washed 3X with PBS and then incubated with Alexa Fluor-conjugated secondary antibodies for 1 hour prior to mounting. Imaging was performed using 40× and 63× objectives on a LSM 510 META confocal microscope (Zeiss).

Immunofluorescent Staining of DIV 4 Mouse Hippocampal Neurons
For preparation of cultured hippocampal neurons, hippocampi were dissected from postnatal day 1 mice and plated at a density of 100,000 cells/well (25). After culture for 4 days in vitro (DIV), neurons were fixed in 4% paraformaldehyde for 10 minutes, washed 4X with PBS, permeabilized, and then blocked overnight in 10% donkey serum + PBS. Cells were stained with goat anti-AKAP150 (Santa Cruz; 1:200) and rabbit anti-Robo2 (Abcam; 1:500) antibodies. Cells were washed 3X with PBS and incubated for one hour at room temperature with Alexa Fluor-conjugated secondary antibodies before mounting. Neurons were imaged using an Axiovert 200M microscope (Zeiss) with a 63× objective (1.4 NA; plan-Apo) and a CoolSNAP2 (Photometrics) CCD camera. Acquisition and off-line processing were conducted using Slidebook 5.5 (Intelligent Imaging Innovations, Denver, CO). Focal plane zstacks (spaced 0.2 µm apart) were acquired and deconvolved to discard out of focus light. Deconvolved 3D z-stacks were then collapsed to generate 2D-maximum-intensity projections.

GST Pulldowns
GST-Robo fusion proteins were purified from E. coli and left on glutathione beads. Mouse brains were homogenized in cold HSE buffer using a Polytron and cleared by centrifuging at 15,000 × g for 30 minutes. Brain lysate or purified protein was added and rocked overnight at 4 degrees. Beads were washed 3X with HSE buffer supplemented to contain 1M NaCl, 2X with regular HSE, and finally with PBS prior to addition of sample buffer. For PP2B pulldowns, lysates were prepared as described above and rocked with glutathione beads for 1 hour at 4 degrees. Beads were washed 3X with buffer containing 150 mM NaCl, 50 mM Tris (pH 7.5), 0.5% NP-40, and protease inhibitors before adding sample buffer. For PKC pulldowns, glutathione beads were rocked overnight with brain lysate and washed with HSE buffer as described above. 100 ng PKCβII was added to beads and incubated in 100 uL buffer containing 150 mM NaCl, 50 mM Tris (pH 7.5), 0.5% NP-40, and protease inhibitors for 1 hour. Samples were washed 3X with the same buffer prior to addition of LDS sample buffer.

RII Overlay
RII overlays were performed as described previously (26,27). In this far-western technique proteins from GST pulldowns were resolved by SDS-PAGE, transferred to nitrocellulose, and blots were subsequently overlaid with digoxigenin-labelled RII subunit. Proteins that bound the labeled RII were identified following incubation with anti-digoxigenin antibody (Abcam).

Immunohistochemistry of Spinal Sections
E11.5 embryos were fixed in 4% paraformaldehyde for 2 hours, washed 4X with PBS, and cryoprotected in 30% sucrose in PBS overnight prior to embedding in OCT. Frozen sections were cut at 20 µm on a cryostat (Leica) and mounted onto Superfrost-plus slides (Fisher Scientific). The sections were blocked and permeabilized for 30 minutes in PBS + 0.1% Triton-X + 1% normal donkey serum (Jackson ImmunoResearch) and stained overnight at 4°C with the following primary antibodies: anti-L1Cam (Chemicon; 1:200), anti-AKAP150 (Santa Cruz; 1:500), and anti-Robo2 (Abcam; 1:500). The slides were then washed 3X in PBS and incubated with the appropriate fluorescent secondary antibody (Jackson ImmunoResearch) for 2 hours at 4°C. Finally, slides were washed 3X in PBS and mounted with Fluromount-G (Southern Biotech). Images were collected using a DMI6000B inverted microscope equipped for both widefield and spinning disc confocal fluorescent microscopy (Leica).

Immunohistochemistry of Sagittal Brain Sections
Brains from 16-week-old wildtype and AKAP150 -/-mice were fixed in formalin for 48 hours prior to paraffin embedding and sagittal sectioning at a thickness of 4 microns by the UW Pathology Research Services lab. Deparaffinized sections underwent antigen retrieval in a pressure cooker before blocking and staining with anti-AKAP150 (V088; 1:1000 and Santa Cruz; 1:500), anti-Robo2 (Abcam; 1:500), or anti-Robo3 (R&D Systems; 1:500) antibodies. The sections were then washed 3X in PBS prior to incubation with Alexa Fluorconjugated secondary antibodies for 1 hour at room temperature.. Sections were imaged using a DMI6000B inverted microscope equipped for both widefield and spinning disc confocal fluorescent microscopy (Leica).

Kinase Assays
For PKA kinase assays, GST-Robo fusion proteins were phosphorylated in kinase assay buffer containing 25 mM Tris pH 7.5, 0.1 mM EGTA, 10 mM MgCl2, 50 µM ATP, and 0.2 µg PKA. PKC kinase assays were performed using 0.2 µg PKC, and the above buffer was supplemented with PKC activation mix (Millipore), 1 µg/mL diacylglycerol, and 10 µg/mL phosphatidylserine. Kinase reactions were stopped by washing 3X with cold HSE buffer prior to addition of LDS sample buffer.

Peptide Array
Peptides corresponding to regions of Robo3 containing potential PKC phosphosites were synthetized using an Auto-Spot Robot ASP 222 as previously described (28). Prior to phosphorylation, membranes were briefly wetted in ethanol and then incubated overnight in preincubation buffer [20 mM HEPES (pH 7.2), 100 mM NaCl, 2mM MgCl 2 , 1 mM EDTA, 1 mM DTT, and 0.2 mg/ml BSA]. Membranes were then blocked for 1 hour at 30 degrees in the same buffer supplemented with 1 mg/mL BSA and 30 µM ATP. Phosphorylation was carried out by incubating membrane in kinase assay buffer [25 mM Tris pH 7.5, 0.1 mM EGTA, 10 mM MgCl2, 50 µM ATP, PKC activation mix (Millipore), 1 µg/mL diacylglycerol, 10 µg/mL phosphatidylserine, and 0.2 µg PKC] for 30 minutes at 30 degrees. Following phosphorylation, membranes were washed 4 times for 15 minutes in 1 M NaCl, 3 times for 5 minutes in H 2 0, 3 times for 15 minutes in 5% phosphoric acid and finally 3 times for 5 minutes in H 2 0. Membranes were blocked for 1 hour in 5% non-fat milk and 1% BSA and incubated overnight with anti-Phospho-Ser PKC Substrate antibody (Cell Signaling).

Proximity Ligation Assay
HEK293 cells were transfected with V5-Robo3, and cultured for 24 hours prior to fixation and permeabilization in buffer containing 20 mM PIPES (pH 6.8), 10 mM EGTA, 1 mM MgCl, 0.2% Triton X-100, and 4% paraformaldehyde. Coverslips were blocked for 1 hour in PBS containing 10% donkey serum and 0.1% fish gelatin (Sigma) and then incubated overnight with anti-Robo3 (R&D Systems; 1:500) and antiphospho-Ser PKC substrates (Cell Signaling; 1:1000) antibodies. The Duolink in situ proximity ligation reaction (Sigma-Aldrich) was carried out according to manufacturer's instructions. To identify transfected cells, coverslips were then incubated overnight with anti-V5 antibody, washed, and incubated for 1 hour at room temperature with Alexa Fluor-conjugated secondary antibody (Invitrogen). Coverslips were mounted using Prolong antifade mounting media containing DAPI (Molecular Probes) and imaged using 63× objective lens on a DMI6000B inverted confocal fluorescent microscope (Leica). The amount of PLA signal for each condition was quantified using Metamorph software (Molecular Devices). Individual cells expressing V5-Robo3 were outlined and the integrated intensities of both the V5-Robo3 and PLA signals were recorded. The average ratio of PLA signal to V5-Robo3 was calculated for each experiment for both control and PDBu treatments. The fold change in normalized PLA/Robo3 ratio was determined and analyzed using a one sample t-test with .05 as the level of significance.
shRNA Knockdown of AKAP79 HEK293 cells were transfected with V5-Robo3, HA-m 1 , and either pSilencer or AKAP79 pSilencer constructs. Cells were incubated for 72 hours posttransfection and then treated with either vehicle (H 2 0) or 10 µM oxotremorine-M for 2 minutes at 37 degrees. Robo3 immunoprecipitations and western blotting were performed as described above. Levels of phospho-Robo3 were determined by densitometric analysis and normalized to total Robo3 expression. Comparisons between pSilencer and AKAP79 pSilencer expressing cells were made using an unpaired t-test with .05 as the level of significance.

Mass spectrometry identifies Robo2 as a putative AKAP79/150 binding partner: Previous work has
shown that AKAP79/150 is a multifunctional anchoring protein that interacts with a range of neuronal binding partners (14,26,29,30). Therefore, we conducted a proteomic screen to look for additional proteins interacting with the murine ortholog, AKAP150. Protein complexes were immunoprecipitated from mouse brain extracts using a polyclonal antibody against AKAP150 and separated by SDS gel electrophoresis ((31); Fig. 1A, lane 2). Control immunoprecipitations were performed with IgG (Fig. 1A, lane 1). Silver stained bands that were enriched in the AKAP150 immunoprecipitation were excised and subjected to protein identification using MALDI-TOF mass spectrometry. Known members of the AKAP150 complex were detected, including the NR2A subunit of the NMDA receptor, thereby validating this approach (18)(19)(20)(21). In addition, we identified several previously unknown AKAP150 binding partners (Fig. 1B). These included key elements of neuronal guidance pathways such as the Roundabout receptor Robo2 and its ligands Slit2 and Slit3 (32,33). These results imply that AKAP150 signaling complexes have the potential to interface with the axon guidance machinery and dendritic ion channels.

AKAP79 interacts with the Robo2 receptor:
In order to validate the results from our proteomics screen, it was important to determine whether AKAP79/150 interacts with Robo2 inside cells. Therefore, HEK293 cells were transfected with plasmids encoding FLAG-tagged AKAP79, the human ortholog of the anchoring protein, and V5tagged Robo2. Cells were harvested after 48 hours and cell lysates were subjected to immunoprecipitation with anti-FLAG antibody. Western blot analysis of AKAP79 immune complexes detected the Robo2 receptor (Fig. 1C, top panel, lane 1). Control experiments performed from cells transfected with only Robo2 plasmid were negative (Fig. 1C, top panel, lane 2). Immunoblot analysis evaluated the expression levels of both proteins in HEK293 cell lysates (Fig.  1C, bottom two panels). Further validation of this protein-protein interaction was provided by reciprocal immunoprecipitation experiments. Immunoblots revealed the presence of AKAP79 in Robo2 immune complexes, but not control immunoprecipitations (Fig. 1D, top panel). Immunoblot analyses evaluated the expression levels of both proteins in HEK293 cell lysates (Fig.  1D, bottom two panels). Collectively, these results allow us to conclude that heterologously expressed AKAP79 and Robo2 interact in HEK293 cells.
Immunofluorescence staining further suggests that Robo2 and AKAP79 reside in the same subcellular compartment (Fig. 1, E-G). Confocal imaging revealed that recombinant AKAP79 (Fig. 1E, red) and Robo2 (Fig. 1F, green) have overlapping regions of expression and accumulate near the plasma membrane in HEK293 cells. This is best observed in the composite image of the merged signals (Fig. 1G). These results imply that AKAP79 and Robo2 co-distribute inside cells.
Robo2 is expressed throughout the developing murine hippocampus, a region of the brain that is enriched with AKAP79/150 (22,34). Thus, a more pertinent validation of the interaction between these two proteins was to be found in primary cultures of mouse hippocampal neurons grown for 15 days in vitro (DIV 15). Endogenous AKAP150 (Fig. 1, H and J, red) was found to overlap with recombinant Robo (Fig. 1, I and J, green). Higher magnification confocal images reveal that both proteins accumulate at the tips of neuronal processes (Fig. 1, K-M, arrows). Further experiments revealed that endogenous AKAP150 (Fig. 1, N and P, red) co-distributed with endogenous Robo2 (Fig. 1, O and P, green) in discrete clusters. These regions of signal overlap were located in both the cell body and dendrites of mouse hippocampal neurons that had been cultured for 4 days. At higher magnification, it was evident that both proteins co-clustered at the tips of neuronal processes, including putative dendritic growth cones (Fig. 1, Q-V).

Mapping the AKAP79-Robo2 binding interface:
On the basis of mass spectrometry identification and biochemical analyses, we postulate that AKAP79/150 and Robo2 can exist as a macromolecular complex. The next phase of these studies was to determine what regions of AKAP79 and Robo2 participate in this protein-protein interaction. To map the Robo2 binding site on AKAP79, we split the anchoring protein into amino-terminal (residues 1-153), central (residues 154-296), and carboxyl-terminal (residues 297-427) fragments. Each AKAP79 fragment was expressed as a GST fusion protein ( Fig. 2A, middle panel). This family of deletion fragments was used to pull down Robo2 from HEK293 cell lysates. Immunoblot analysis revealed that Robo2 selectively bound to the amino-terminal fragment of AKAP79, suggesting that its principal binding site lies between residues 1-153 of the anchoring protein ( Fig. 2A, top panel, lane 2). It should be noted that Robo2 weakly interacted with the carboxyl terminal regions of the anchoring protein ( Fig. 2A, top panel, lane 4). Control experiments confirmed that equivalent amounts of Robo2 were used in all binding experiments ( Fig. 2A, bottom  panel). These experiments suggest that multiple regions of AKAP79 may interface with Robo2.
In reciprocal experiments, a similar approach was used to map the region of Robo2 that binds to the anchoring protein. The cytoplasmic tail of Robo2 was split into three fragments (Fig. 2B, middle  panel). These GST-fusion proteins were then used to pull down FLAG-AKAP79 from HEK293 cell lysate. The anchoring protein bound strongly to a central portion of Robo2 that encompasses residues 991-1141 (Fig. 2B, top panel, lane 3), although weaker binding of AKAP79 was also detected to a fragment of Robo2 that is immediately proximal to the membrane-spanning segment (residues 881 to 990; Fig. 2B, top panel,  lane 2). This led us to conclude that the primary AKAP79 binding site lies between residues 991-1141 of Robo2. Control experiments confirmed that equivalent amounts of AKAP79 were used in all binding experiments (Fig. 2B, bottom panel).
To further narrow down the AKAP79 binding sites, we generated an additional family of truncations in the cytoplasmic tail of Robo2 (Fig. 2C). AKAP79 binding was analyzed as described above. These studies revealed that a Robo2 receptor expressing the first 1070 residues was still competent to bind AKAP79. This represents a region of approximately 190 residues that reside in the cytoplasm of cells (Fig. 2C). Further truncation of the cytoplasmic tail of Robo2 prevented interaction with the anchoring protein (Fig. 2D,  top panel, lane 5). Thus, we can conclude that residues 991-1070 of Robo2 are necessary for interaction with AKAP79. Control experiments confirmed that equivalent amounts of Robo2 and AKAP79 were used in all binding experiments (Fig. 2D, middle and bottom panels).

AKAP79/150 binds directly to the cytoplasmic tails of Robo2 and Robo3
: Studies in the previous sections indicate that human AKAP79 and its murine ortholog AKAP150 have the capacity to interact with Robo2. Next, we investigated whether Robo2 is capable of interacting with endogenous AKAP150 and its anchored enzymes. A GST-tagged Robo2 881-1378 fragment was used to perform pulldowns from mouse brain lysate. Overlay with labeled RII revealed that an AKAP of the molecular weight of AKAP150 was bound to this Robo2 fragment (Fig. 2E, upper  panel, lane 2). Western blotting confirmed that this band corresponded to endogenous AKAP150 (Fig. 2E, top panel, lane 2). The anchoring protein was not detected in control pull downs using GST alone (Fig. 2E, top two panels, lane 1). Analysis of mouse brain lysates verified that equivalent amounts of AKAP150 were present in both samples (Fig. 2E, lower two panels). Related experiments established that the AKAP79/150 binding partners RIIα, the phosphatase PP2B (B subunit), and conventional PKCs co-purify with the Robo2 881-1378 fragment (Fig. 2F, top three  panels, lane 2). Collectively, these findings strengthen our view that Robo2 associates with AKAP150 complexes containing a range of anchored signaling enzymes. Additional experiments revealed that GST-Robo2 also binds human AKAP79 from HEK293 cell lysates (Fig.  2G, top panel, lane 4).
AKAP79/150 binds directly to certain membrane proteins, such as the L-type calcium channel and the KCNQ2 subunits of the M-channel (22,30,(35)(36)(37)(38). In contrast, NMDA and AMPA receptors interact with AKAP79/150 through a bridging interaction with membrane-associated guanylate kinase (MAGUK) proteins (18,19,39). To determine if AKAP79 can interface directly with Robo2, we repeated our GST-Robo2 pulldowns using His-tagged AKAP79 purified from E. coli. These experiments confirmed that AKAP79 binds directly to Robo2, without the need for an adaptor protein (Fig. 2H, top panel, lane 4). Purified recombinant RII was used as a negative control for these binding studies (Fig. 2H, bottom panel, lane  2).
There are four members of the Robo family of chemotactic guidance receptors (40). Therefore, it seemed logical to explore the possibility that AKAP79/150 may associate with other Robo receptor isoforms. However, only Robo1, 2, and 3 are expressed in neurons (40). Therefore, AKAP79 pulldown experiments were conducted with GSTfusion proteins encompassing the cytoplasmic tails of Robo1, Robo2, and Robo3. These binding studies revealed that Robo3 also associates with AKAP79 (Fig. 2I, top panel, lane 4). GST-Robo2 binding to AKAP79 served as an internal control (Fig. 2I, top panel, lane 3). Immunoblot analysis confirmed that equivalent amounts of each binding partner were present in these experiments (Fig. 2I, middle and lower panels). Sequence homology between Robo2 and Robo3 within the region that spans the AKAP binding site may explain why the AKAP is capable of binding both of these receptors. In contrast, the Robo1 isoform contains an insertion within this region, potentially explaining why it does not associate with this anchoring protein (Fig. 2I, top panel, lane 2).
Much of the knowledge of Roundabout receptors has been gleaned from the investigation of axonal guidance in the developing spinal cord. Signals processed by Robo1 and Robo2 expel axons from the floor plate of the spinal cord and prevent inappropriate recrossing of the midline (41). In contrast, signaling through Robo3 promotes midline crossing (42). Therefore, we reasoned that association with AKAP150 might influence aspects of axonal guidance in mouse spinal commissural axons. As a prelude to these studies, we examined whether AKAP150 is expressed in the developing spinal cord of E11.5 mice. Immunostaining revealed that AKAP150 is broadly expressed throughout the developing spinal cord (Fig. 3A, green). In keeping with published reports, Robo2 expression was restricted to the ventral and lateral funiculi ((43) ; Fig. 3, B and C, red). These two regions of the spinal cord were re-examined at higher magnification by confocal microscopy (Fig. 3, D-J). Co-distribution of both signals was most apparent in the ventral and lateral funiculi (Fig. 3, D-F, yellow), but some overlap of staining was also observed in the cell bodies of commissural neurons (Fig. 3, H-J,  yellow). Control experiments using the same antibodies confirmed that the AKAP150 signal was absent in equivalent sections from E11.5 AKAP150 -/-mice ((22) Fig. 3, K, N, and Q). Robo2 staining appeared normal in these sections (Fig. 3, L-P). Normal patterns of midline crossing were observed when wildtype and AKAP150 -/sections were stained for neural cell adhesion molecule L1, a post-crossing axonal marker ((44) ;  Fig. 3, T and U). Thus, it would appear that although AKAP150 and Robo2 are co-distributed in the developing spinal cord, their association does not influence midline crossing of spinal commissural axons in mouse embryos.

Robo2 and Robo3 co-distribute with AKAP150 in the adult mouse brain:
Robo receptors are expressed in the adult brain (45). Likewise, neuronal AKAP150 expression peaks later in development (20). Therefore, we wondered if AKAP150 co-distributed with Robo receptors in the brains of mature mice. To test this possibility, brains from 4-month-old wildtype and AKAP150 -/-mice were isolated and paraffin embedded. Sagittal sections were prepared and immunofluorescent staining with antibodies to AKAP150, Robo2, and Robo3 was completed. Fluorescent microscopy of intact sagittal sections at low magnification revealed distinct but overlapping staining patterns for AKAP150 and Robo2 and Robo3 in a variety of brain regions (data not shown). Widefield fluorescent images (40× objective) show overlapping expression patterns for AKAP150 (red) with both Roundabout receptors (green) in hippocampal sections (Fig. 4, A and E). Upon higher magnification (63×) analysis on a confocal microscope, the co-distribution of the anchoring protein with Robo2 (Fig. 4, B-D) and Robo3 (Fig.  4, F-H) was evident in the CA1 region of the hippocampus. Likewise, analysis of lower magnification images detected both Roundabout receptors co-distributed with AKAP150 in the islands of calleja in the olfactory tubercle (Fig. 4, I  and M, yellow). Overlapping staining patterns between AKAP150 and both Robo2 and Robo3 were also evident in this same tissue at higher magnification (Fig. 4, J-L and N-P). Staining patterns of Robo2 and Robo3 were unchanged in brain sections from AKAP150 -/-mice (data not shown).

Robo3 is phosphorylated by anchored PKC:
AKAP150 expression in the hippocampus peaks at postnatal day 14 in mice (20), a developmental stage when Robo receptors, their Slit ligands and other axon guidance molecules are also present in the hippocampus (43). For instance, the Netrin receptor Deleted in Colorectal Cancer is enriched in the postsynaptic density where it is required to facilitate long-term potentiation, an electrophysiological paradigm for learning and memory (46). Therefore, it seemed plausible that Robo receptors might contribute to acute neuronal signaling events that control synaptic strength. An attractive feature of this hypothesis is that AKAP79/150 signaling complexes also populate the postsynaptic density where they provide local control of synaptic strength (31,47). In keeping with this notion, deletion of AKAP150 in mice impairs long-term depression due to the loss of PKA, PKC, and PP2B anchoring (22,24,25).
In light of these findings, we performed in vitro kinase assays to determine if any Robo receptors are targets of AKAP150-anchored kinases. Purified GST-tagged cytoplasmic tails of Robo1, Robo2, and Robo3 were immobilized on beads and phosphorylated in vitro with a mixture of conventional PKCs and [ 32 P] ATP. Autoradiography revealed that only Robo3 was phosphorylated by PKC (Fig. 5A, top panel, lane  3). Analogous experiments showed that Robo1 was phosphorylated by protein kinase A whereas Robo3 was labeled to a lesser extent (Fig 5B, top  panel, lane 2 and 8). However, we were unable to recapitulate PKA phosphorylation of either Robo receptor inside cells. Since Slit/Robo signaling can mobilize intracellular calcium (48), we reasoned that the selective phosphorylation of Robo3 by PKC might contribute to the unique function of this receptor. For these reasons, we focused on investigating PKC phosphorylation of Robo3.
Next, we performed experiments in the presence of AKAP79 to investigate the role of the anchoring protein in this phosphorylation event. Formation of a Robo3-AKAP79-PKC ternary complex enhanced phosphorylation of the substrate (Fig. 5C, top panel, lane 2) as compared to experiments performed in the absence of the anchoring protein (Fig. 5C, top panel, lane 1). This suggests that AKAP79-anchored PKC preferentially phosphorylates Robo3. A more stringent validation of this concept was performed in HEK293 cells as they express high levels of endogenous AKAP79. Cells were transfected with plasmids encoding full length V5 tagged-Robo3 and PKC activity was stimulated upon application of phorbol 12,13-dibutyrate (PDBu). Robo3 immune complexes were probed with an antibody that detects phosphorylated PKC substrates. Immunoblot analysis revealed robust phorbol ester-dependent phosphorylation of Robo3 (Fig.  5D, top panel, lane 2) as compared to unstimulated controls (Fig. 5D, top panel, lane 1). Immunoblot analysis of whole cell lysates using the phospho-PKC substrates antibody confirmed kinase activation, and additional controls established that equivalent levels of Robo3 were present in each sample (Fig. 5D, middle and  bottom panels).
We subsequently probed a family of Robo3 truncation fragments in an attempt to locate the sites of PKC phosphorylation. These mapping studies indicate that the extreme carboxyl terminus of Robo3 (residues 1236-1386) contained the site(s) of PKC phosphorylation (Fig. 5E, top panel,  lane 2). This region of the Robo3 gene is subject to differential splicing and results in the expression of two subtypes, Robo3.1 and Robo3.2 (Fig. 5F). Plasmids encoding both Robo3 subtypes were expressed in HEK293 cells (Fig. 5G). Cell based phosphorylation experiments demonstrate that only the Robo3.1 form is phosphorylated by PKC inside cells (Fig. 5G, top panel, lane 1).
Unfortunately, quantitative analysis by mass spectrometry was unsuccessful in obtaining more precise information about the identity of these putative phosphorylation sites. Therefore, we utilized peptide spot arrays as an alternate means to approach this problem (Fig. 5, H-L). There are twenty serine or threonine residues in the unique cytoplasmic region of Robo3.1 (Fig. 5H). Given the number of possible phosphosite combinations, we focused on three segments of Robo3.1 sequence that contain consensus PKC substrate motifs (peptides A, B and C are denoted in Fig. 5, H and I). A total of sixteen immobilized peptides were evaluated (Fig. 5, I-K). These included wildtype sequences and peptides where alanine was substituted for putative target serines (Fig. 5I). UV analysis confirmed equivalent spotting efficiency for all peptides (Fig 5L).
In vitro phosphorylation of wildtype peptides with purified PKC revealed that only peptide A was labeled (Fig 5I). This peptide contains Ser1330, which conforms to a consensus PKC substrate site (Fig 5M). Importantly, an alanine-substituted peptide analog of this sequence was not phosphorylated by this kinase (Fig. 5, I-K). Cellbased validation of this result was provided by analysis of a Robo3.1 S1330A mutant. As expected, wildtype Robo3.1 was efficiently phosphorylated upon stimulation of PKC (Fig. 5N,  top panel, lane 2). In contrast, the Robo3.1 S1330A mutant was not phosphorylated by phorbol ester-mediated stimulation of endogenous PKC (Fig. 5N, top panel, lane 4). Loading controls confirmed equal expression of both Robo3.1 forms and immunoblot analysis using phospho-substrate antibodies detected phorbol ester stimulation of PKCs (Fig. 5N, middle and bottom panels). Taken together the data in figure 5 identifies Ser1330 on Robo3.1 as a target for PKC in vitro and inside cells.
Cellular analysis of Robo3.1 phosphorylation was further examined by proximity ligation assay (PLA), a sensitive method that detects proteinprotein interactions or local posttranslational modification of proteins in situ (49). The conceptual basis of this approach depends on the dual proximal binding of distinct antibody probes that are conjugated to complementary oligonucleotides. When the probes are within range of each other, ligation occurs to generate an amplifiable DNA circle. The resulting incorporation of fluorescent nucleotides following multiple rounds of amplification serves as a visible marker for protein-protein interactions or covalent modification of proteins that occur within a radius of 40 to 60 nm. Cells were transfected with Robo3.1 and stimulated with PDBu to activate PKCs. Proximity ligation allowed us to use the phospho-PKC substrates antibody to selectively monitor PKC activity in the vicinity of Robo3.1. Accordingly we were able to show that Robo3.1 was efficiently phosphorylated by PKC (Fig. 6, A-C and G) as compared to unstimulated controls (Fig. 6, D-F and G). Quantitation of this result is shown as the ratio of PLA signal to the single stain for Robo3.1 (Fig. 6G).
Finally, gene silencing was utilized to discern a role for the anchoring protein in the phosphorylation of Robo3.1 (Fig. 6, H and I). A well-characterized shRNA to AKAP79 was used to deplete the anchoring protein in HEK293 cells ((24) : Fig 6H, lower middle panel, lanes 3 and 4). In order to accentuate the physiological relevance of this anchored phosphorylation event, we used the m 1 muscarinic receptor agonist oxotremorine-M (Oxo-M) as an activator of anchored PKC (50). Cells expressing both Robo3 and the m 1 receptor and transfected with either control or AKAP79 shRNA were incubated for 72 h. Following treatment with vehicle or Oxo-M, Robo3 immune complexes were isolated. Phospho-Robo3.1 levels were evaluated by immunoblotting (Fig. 6H, top  panel, lanes 2 and 4). Analysis of data from four independent experiments showed that depletion of AKAP79 significantly reduced PKC phosphorylation of Robo3.1 as assessed by densitometry (Fig. 6I, columns 2 and 4). Overall the data in figure 6 demonstrates that AKAP79 facilitates PKC phosphorylation of Robo3.1 in response to a physiological agonist.

DISCUSSION
A-kinase-anchoring proteins modulate membrane proximal signaling events, principally by tethering protein kinases and phosphatases in the vicinity of G-protein coupled receptors, ion channels, and adenylyl cyclases (7,25,(51)(52)(53)(54). Taken together the data in figure 1 imply that an additional role for AKAPs may be to provide a molecular interface between signaling enzymes and the axon guidance machinery. Axon guidance receptors, including the Robo family, are transmembrane proteins that are enriched in the growth cones of migrating axons (55). However, it has also been shown that Robo family members are expressed in the forebrain, hippocampus and other regions of the developing brain (34,45).
Certain pools of guidance receptors can associate with lipid rafts where they respond to chemotactic signals that instruct directional movement of axons (56). Interestingly, these guidance cues can either attract or repel axons (57). For example, binding of the Slit ligand is a repulsive cue that causes the growth cone to collapse and reverses the direction of the migrating axon (34). Robo receptors are also expressed in dendrites, where Slit binding has been shown to regulate dendritic branching (58). Likewise, the Ephrin and Plexin classes of guidance receptors have been implicated in growth cone repulsion and the propagation of dendritic arborization (59). It is also worthy to note that second messenger signaling events impact neuronal migration. Classic studies have shown that altering the levels of cyclic nucleotides in the growth cone can interfere with axon guidance signals (60). For example, increasing the ratio of cAMP/cGMP can switch the action of guidance molecules from neuronal repulsion to attraction (61). Added to this, it has been demonstrated that elevation of intracellular cAMP and the subsequent activation of PKA can suppress Robo signaling (62).
Kinase anchoring contributes to various modes of directional neurite outgrowth control (10). For example, the AKAP WAVE1 is expressed at the leading edge of growth cones, where it controls actin dynamics to facilitate axonal protrusions (63)(64)(65)(66). In addition, the Drosophila anchoring protein Nervy coordinates signaling at the Plexin family of Semaphorin receptors (67). Nervy is purported to cluster PKA at the Plexin receptor, thereby facilitating the termination of Semaphorin signaling through the phosphorylation of these receptors (67). Taken together these studies provide a precedent for the notion that local signaling mechanisms govern dynamic aspects of neuronal guidance and development. The data presented in this report provides the first evidence that AKAP-associated enzymes may regulate Robo receptors. We propose that the interaction of AKAP79/150 with Robo receptors may position PKC for a role in the fine-tuning of axonal and dendritic guidance.
The accumulated data presented in figure 2 demonstrate that the cytoplasmic tail of Robo receptors primarily interfaces with the aminoterminal region of AKAP79/150. Previous studies have shown that this segment of the anchoring protein contains three polybasic regions that participate in membrane targeting through interaction with negatively charged phospholipids (29). More recent studies have implicated the palmitoylation of conserved cysteines in the proximal and distal polybasic regions as a means to guide AKAP79/150 to lipid rafts (21,68). These latter findings raise the intriguing possibility that AKAP-Robo2 assemblies might be sequestered in specialized sub-membrane compartments. This may be particularly relevant in regions of the hippocampus where AKAP79/150 appears to codistribute with Robo2 and Robo3, particularly at the growth cones of developing neurons. Although the immunofluorescence images presented in figures 1 and 4 are consistent with this notion, super-resolution approaches will be necessary to locate precisely where AKAP-Robo signaling units are sequestered in neuronal membranes. In addition, it remains possible that ancillary proteinprotein or protein-lipid interactions may further secure AKAP-Robo macromolecular assemblies at synaptic locations.
We also demonstrate that AKAP79-bound signaling enzymes such as PKA, PKC, and PP2B can remain associated with the anchoring protein when in complex with Robo receptors. The data presented in figure 2F suggests that kinase and phosphatase anchoring may coordinate the reversible phosphorylation of Robo receptors in a manner analogous to the GluA1 subunit of the AMPA receptor (19). Thus, several distinct targeting mechanisms may act synergistically with the AKAP-Robo interface to stably sequester AKAP-associated signaling enzymes with these transmembrane receptors.
Protein kinase C requires basic residues in the vicinity of the phosphorylated amino acid, but its specificity is less well defined than other kinases such as PKA or Akt (69). Taken together, the results from figure 5 suggest that AKAP79 directs PKC proximal to Robo3.1 to enhance phosphorylation of its cytoplasmic tail. Analysis of peptide spot arrays when combined with sitedirected mutagenesis has identified Ser1330 as a primary site of PKC phosphorylation. This region is rich in basic and bulky hydrophobic residues, hallmarks of a consensus site for PKCs (69). Data from figure 6 confirms and extends these observations by demonstrating that AKAP79 facilitates the phosphorylation of Robo3.1 by PKC. Perhaps the strongest support for this latter statement is provided by the gene silencing experiments presented in figures 6H and I. Depletion of the anchoring protein suppresses phosphorylation of Robo3.1 in response to muscarinic agonists. Although little is known about Robo3 receptor subtypes there is reason to believe that Robo3.1 and Robo3.2 have divergent functions. In the developing spinal cord, Robo3.1 antagonizes repulsive axonal migration events that proceed through Robo1 and Robo2 receptors (70). This ability to repress canonical Slit/Robo signals is not shared by the Robo3.2 isoform. While the mechanism that accounts for the opposing functions of Robo3.1 and Robo3.2 is not clear, our data raise the possibility that PKC phosphorylation at Ser1330 of Robo3.1 may play a role in this process.
In conclusion, our data suggests that PKC anchoring through AKAP79 may confer an added element of spatiotemporal control to Slit/Robo signaling. AKAP79 can be thought of as an adapter molecule that brings enzyme and substrate together. This mechanism not only provides exquisite local modulation of Robo3.1 phosphorylation, but the immediate proximity of each component ensures that signals have the capacity to be instantaneously transduced. Furthermore, AKAP79/150 has the ability to bind all isoforms of PKC, which are activated by different combinations of calcium and phospholipid (8). This anchoring protein can also self-associate and complex with ion channels (17,71). Finally, Robo receptor subtypes homo-and hetero-dimerize (72). These factors suggest that incorporation of the PKC-AKAP-Robo3.1 subcomplexes into higher order macromolecular assemblies may expand the potential for signal integration and enzyme crosstalk between second messenger cascades and the neuronal guidance machinery.
membrane-associated guanylate kinase; PDBu, phorbol 12,13-dibutyrate; PLA, proximity ligation assay: Oxo-M, oxotremorine-M.   4) were used to pull down overexpressed AKAP79 from HEK293 cell lysates. C, diagram depicting progressive truncation of the Robo2 receptor used for fine mapping of the AKAP79 binding site. The first and last residues of each fragment are denoted. D, AKAP79 and truncated Robo2 receptors were overexpressed in HEK293 cells. AKAP79 immune complexes were immunoprecipitated and Robo2 binding was assessed by immunoblot. E, purified Robo2 C-terminus (Robo2-CT) fused to GST was incubated with mouse brain lysate. Immunoblots were probed for AKAP150 (top) and overlaid with digoxigenin-labeled RII (upper middle). F, GST-Robo2CT pulldowns from mouse brain lysate were immunoblotted for RIIa (top), PP2B (second from top), and PKC (third from top). G, GST-Robo2CT pulldowns from HEK293 cell lysate containing overexpressed AKAP79 were probed for AKAP79 (top). H, GST-Robo2CT was incubated with either purified His-AKAP79 or His-RII. GST-Robo2CT complexes were immunoblotted for AKAP79 (top) and RII (upper middle). I, the C-termini of Robo1, Robo2 and Robo3 were each fused to GST and purified from bacteria. Purified His-AKAP79 was incubated with the GST-RoboCT proteins and immunoblots of the GST-Robo complexes were probed for AKAP79 (top). Figure 3. AKAP150 is expressed in the developing spinal cord where it co-distributes with Robo2 in specialized regions. Transverse sections of E11.5 spinal cords from wildtype and AKAP150 -/-mice were stained for AKAP150 and Robo2, and imaged using widefield fluorescence microscopy (wildtype; A-C and AKAP150 -/-; K-M). AKAP150 is shown in green and Robo2 is shown in red. Higher magnification spinning disc confocal images provide more detailed localization of these proteins in the funiculi (wildtype; D-F and AKAP150 -/-N-P) and commissural neuron cell bodies (wildtype; H-J and Figure 1 by guest on July 9, 2020