cAMP at Perinuclear mAKAPα Signalosomes Is Regulated by Local Ca2+ Signaling in Primary Hippocampal Neurons

Abstract The second messenger cyclic adenosine monophosphate (cAMP) is important for the regulation of neuronal structure and function, including neurite extension. A perinuclear cAMP compartment organized by the scaffold protein muscle A-kinase anchoring protein α (mAKAPα/AKAP6α) is sufficient and necessary for axon growth by rat hippocampal neurons in vitro. Here, we report that cAMP at mAKAPα signalosomes is regulated by local Ca2+ signaling that mediates activity-dependent cAMP elevation within that compartment. Simultaneous Forster resonance energy transfer (FRET) imaging using the protein kinase A (PKA) activity reporter AKAR4 and intensiometric imaging using the RCaMP1h fluorescent Ca2+ sensor revealed that membrane depolarization by KCl selectively induced activation of perinuclear PKA activity. Activity-dependent perinuclear PKA activity was dependent on expression of the mAKAPα scaffold, while both perinuclear Ca2+ elevation and PKA activation were dependent on voltage-dependent L-type Ca2+ channel activity. Importantly, chelation of Ca2+ by a nuclear envelope-localized parvalbumin fusion protein inhibited both activity-induced perinuclear PKA activity and axon elongation. Together, this study provides evidence for a model in which a neuronal perinuclear cAMP compartment is locally regulated by activity-dependent Ca2+ influx, providing local control for the enhancement of neurite extension.


Introduction
CNS neurons responsible for higher order functions fail to survive or regenerate their axons after injury, resulting in permanent disability in common diseases such as stroke, Alzheimer's disease, Parkinson's disease and glaucoma. To combat this disability, strategies are being sought to promote CNS neuron survival and axon regeneration after injury, including the identification of intracellular signaling pathways whose activation might be beneficial in disease. Enhanced cyclic adenosine monophosphate (cAMP) signaling has been shown to potentiate neurotrophic signaling and to promote neuron survival and axon regeneration (Wang et al., 2015;Wild and Dell'Acqua, 2018). cAMP associated with these processes can be activity dependent (Goldberg and Barres, 2000;Goldberg et al., 2002;Corredor et al., 2012), but the mechanisms conferring this regulation remain largely unknown.
Although cAMP is in theory a freely diffusible second messenger present throughout the cell, it is now established that the specific effects of cAMP signaling in response to different stimuli often occur in discrete intracellular compartments organized by scaffold proteins that form multimolecular signaling complexes or "signalosomes" (Wild and Dell'Acqua, 2018). Scaffolds that bind the cAMP effector protein kinase A (PKA) are called A-kinase anchoring proteins (AKAPs). Diverse neuronal functions, including synaptic plasticity, neuronal excitability and transduction of sensory information, have been shown to be associated with AKAP-mediated compartmentation (Wild and Dell'Acqua, 2018). Recent studies have implicated muscle AKAPa (mAKAPa) in prosurvival and progrowth neurotropic and cAMP signal transduction, including in the extension of neurites by hippocampal and retinal neurons in vitro (Wang et al., 2015;Boczek et al., 2019).
Expressed in neurons and striated myocytes, the large 250-kDa mAKAP (AKAP6) scaffold (a-isoform in neurons, b -isoform in myocytes) is localized to the nuclear envelope via binding to the Klarsicht/ANC-1/Syne-1 homology (KASH) domain, transmembrane protein nesprin-1a (Pare et al., 2005a;Boczek et al., 2019). mAKAP binds .20 different signaling enzymes and gene regulatory proteins, thereby regulating stress-induced gene expression in these excitable cells (Dodge-Kafka et al., 2019). mAKAP was the first AKAP to be shown to be capable of binding an adenylyl cyclase (AC), a phosphodiesterase (PDE), and a cAMP effector, thus having the potential to orchestrate completely compartmentalized cAMP signaling (Dodge et al., 2001;Dodge-Kafka et al., 2005;Kapiloff et al., 2009). By expression of nesprin-1a-localized constitutive active AC and PDE fusion proteins, cAMP at mAKAPa signalosomes has been shown to be sufficient and necessary for neurite extension by embryonic day (E)18 rat hippocampal neurons in vitro (Boczek et al., 2019). Inhibition of local cAMP signaling by the PDE-nesprin-1a fusion protein both suppressed forskolin-induced PKA activity detected by a nuclear envelope-localized Forster resonance energy transfer (FRET) PKA reporter (PN-AKAR4) and blocked activity-dependent neurite extension. In contrast, expression of the AC-nesprin-1a fusion protein that increased cAMP levels at mAKAPa signalosomes promoted neurite outgrowth. In addition, anchoring disruptor peptide-mediated displacement of endogenous type 4D3 PDE from mAKAPa signalosomes similarly elevated perinuclear cAMP levels and potentiated neurite extension. mAKAPa signalosomes have been implicated not only in the regulation of neurite extension, but also prosurvival signaling. PDE displacement enhanced retinal ganglion survival in vivo after optic nerve crush, consistent with prior findings that the mAKAPa scaffold is required for the neuroprotective effects of exogenous cAMP after crush injury (Wang et al., 2015;Boczek et al., 2019). Using live cell imaging, we now consider the activitydependent regulation of cAMP levels at mAKAPa signalosomes in neurons, demonstrating local production of cAMP in that compartment and local regulation of neurite extension.

Hippocampal neuron isolation and culture
All procedures for animal handling were approved by the Institutional Animal Care and Use Committee at Stanford University. Primary hippocampal neurons were isolated from E18 Sprague Dawley rat embryos of either sex. Briefly, the hippocampal CA1-CA3 region was dissected in PBS medium with 10 mM D-glucose and digested with 0.05% trypsin-EDTA in PBS with 11 mM Dglucose for 30 min at 37°C. The dissociated tissues were centrifuged at 250 Â g for 2 min and then triturated with fire polished glass pipet in HBSS with calcium and magnesium in plating medium (10% v/v horse serum in DMEM). Dissociated neurons were plated on nitric acidtreated 25 mm cover glass coated with poly-L-lysine in plating medium. Four hours after plating, the medium was replaced with maintenance Neurobasal defined medium supplemented with 1% N2, 2% B27 (Invitrogen), 5 mM D-glucose, 1 mM sodium pyruvate. On days 4-5 in culture, 4 mM arabinosyl cytosine was added to inhibit glial proliferation, and the neurons were plasmid transfected with Lipofectamine 3000 and/or infected with adenovirus.

Neurite extension assays
For neurite extension assays, the cells were cultured for 3 d in maintenance Neurobasal defined medium supplemented with 1% N2, 2% B27 (Invitrogen), and 1 mM sodium pyruvate. On days 3-4 in culture, 4 mM arabinosyl cytosine was added to inhibit glial proliferation, and on day 4, the neurons were co-transfected with pmCherry-C1 and Parv-GFP-nesprin or control GFP-nesprin expression plasmids using Lipofectamine LTX with Plus Reagent (ThermoFisher Scientific, catalog #15338030). KCl (40 mM) was added to the medium after transfection. Two days later, the neurons were fixed and counterstained with Hoechst (Invitrogen, catalog #33342). Images were acquired with a Zeiss 880 confocal microscope by 20Â objective tile scan and processed with Fiji ImageJ. The length of the longest neurite for 20-40 neurons per condition was measured for each experiment using ImageJ with the Simple Neurite Tracer plugin.

Statistical analysis
Statistical analyses were performed using GraphPad Prism. Data are presented as mean 6 SEM. Normally distributed datasets by D'Agostino-Pearson omnibus (K2) test were compared by unpaired t tests (for two groups) or one-way ANOVA (for three groups) with subsequent Tukey's post hoc testing. Other datasets were analyzed by Mann-Whitney U test (for two groups) or Kruskal-Wallis H test (for three groups), followed by Dunn's post hoc testing. Repeated symbols are used as follows: single, p 0.05; double, p 0.01; triple, p 0.001.

Results
KCl depolarization induces PKA and Ca 21 transients at the nuclear envelope cAMP-dependent PKA signaling is highly compartmentalized in cells by AKAPs that organize localized signalosomes regulating specific cellular processes (Wild and Dell'Acqua, 2018). Given that perinuclear cAMP signaling has been shown to be required for activity-dependent neurite extension (Boczek et al., 2019), we now considered whether KCl-mediated depolarization can regulate cAMP at mAKAPa signalosomes. Cultured primary E18 rat hippocampal neurons were transfected on days 4-5 in culture with expression plasmids for perinuclear-localized PN-AKAR4 or diffusely localized parent AKAR4 PKA activity FRET biosensors (Fig. 1A). At the same time, the neurons were co-transfected with plasmids expressing perinuclear-localized PN-RCaMP1h or diffusely localized parent RCaMP1h intensiometric Ca 21 sensors (Fig. 1A). The neurons were imaged 36-72 h after transfection. As AKAR4 emits cyan and yellow light, while RCaMP1h is red, we were able to image simultaneously similarly localized PKA and Ca 21 sensors.
Membrane depolarization with 40 mM KCl for 60 s induced a pronounced increase in perinuclear PKA activity, but had no significant effect on PKA detected with the diffusely localized PKA parent sensor ( Fig. 1B-D). In addition, whereas 40 mM KCl resulted in robust perinuclear PKA activation, 10 mM KCl did not induce activation of perinuclear PKA, and 30 mM KCl inconsistently resulted in PN-AKAR4 signals (Fig. 1E,F). This was in contrast to the similarly robust response by AKAR4 and PN-AKAR4 to the transmembrane AC activator forskolin previously observed in these neurons (Boczek et al., 2019). KCl induced Ca 21 transients in both compartments, notably with less increase in RcAMP1h signal at the nuclear envelope (Fig.  1C,D). These results imply KCl depolarization can selectively activate PKA at the nuclear envelope, despite elevating [Ca 21 ] more generally in the cell. mAKAPa is required for activity-induced perinuclear PKA signaling As PKA is recruited to the nesprin-1a perinuclear compartment by the scaffold mAKAPa, it was likely that KClinduced PN-AKAR4 signal would be because of activation of mAKAPa-bound PKA. Expression of a shRNA that has been used previously to deplete cells of mAKAP (Pare et al., 2005b;Boczek et al., 2019) inhibited KCl-induced perinuclear PKA activity ( Fig. 2A,B). In contrast, mAKAPa expression was not required for KCl-induced Ca 21 Figure 1. Depolarization selectively activates PKA signaling at the nuclear envelope. A, Sensors used in this study. In AKAR4, phosphorylation of the LRRATLVD peptide by PKA results in FHA1 phospho-peptide binding and increased cerulean-cpVenus FRET (Depry et al., 2011). In RCaMP1h, Ca 21 induces the binding of the M13 peptide by the mutant calmodulin domain (mCaM), increasing mRuby fluorescence (Hirabayashi et al., 2017). In the perinuclear-localized sensors PN-AKAR4 and PN-RCaMP1h, nesprin-1a contains five spectrin repeats (SRs) and a transmembrane KASH domain that localizes the protein to the nuclear envelope via binding to SUN domain proteins (Pare et al., 2005a). B, Grayscale CFP images of hippocampal neurons expressing AKAR4 or PN-AKAR4. Scale bar: 10 mm. C, D, Hippocampal neurons were transfected with PN-AKAR4 and PN-RCaMP1h or AKAR4 and RCaMP1h expression plasmids. Representative traces (smoothed with Prism) and pseudocolor images showing FRET (AKAR4) or intensity (RCaMP1h) responses to 40 mM KCl introduced by perfusion. Images were obtained simultaneously for AKAR4 and transients at the nuclear envelope, such that mAKAP depletion had no effect on Ca 21 transients detected with the PN-RCaMP1h sensor (Fig. 2C,D). Together, these data are consistent with a model in which mAKAPa is required for the recruitment of PKA to the mAKAPa-nesprin-1a perinuclear compartment, but not for the release of Ca 21 into that compartment.
Perinuclear Ca 21 dynamics and PKA activity depends on L-type Ca 21 channel activity We investigated which voltage-gated channels might contribute to Ca 21 fluxes in the mAKAPa perinuclear compartment. Preincubation of hippocampal neurons with the L-type Ca 21 channel blocker nifedipine inhibited the Ca 21 transients detected by the parent diffusely localized RCaMP1h sensor 51% in amplitude and that detected by the PN-RCaMP1h sensor 92% in amplitude (Fig. 3A,B,E,F). Preincubation with the N-type Ca 21 channel blocker conotoxin GVIA inhibited the Ca 21 transients detected by the parent diffusely localized RCaMP1h sensor 55% in amplitude and that detected by the PN-RCaMP1h sensor 60% in amplitude. Accordingly, nifedipine, but not conotoxin GVIA, significantly inhibited KCldependent PKA transients detected by the PN-AKAR4-1a sensor (Fig. 3C,D). Preincubation of the neurons with Ca 21 channel blockers did not alter the lack of response of the parent AKAR4 sensor to KCl depolarization (Fig.  3G,H). Taken together, these results suggest that Ca 21dependent PKA activity in the mAKAPa-nesprin-1a compartment is preferentially dependent on L-type channel activity.
Chelation of calcium at the nuclear envelope inhibits activity-induced cAMP change As KCl-mediated neuronal depolarization activated perinuclear PKA via an L-type Ca 21 channel-dependent mechanism, we next considered whether the Ca 21 influx promoting PKA activity was local to mAKAPa signalosomes or elsewhere in the cell besides that compartment. Carp parvalbumin-b is a high affinity (K a = 29 nM) Ca 21binding protein with 10 4 -fold Ca 21 selectivity over Mg 21 (Wang et al., 2013). To deplete the perinuclear compartment of Ca 21 , we expressed a parvalbumin-b -nesprin-1a fusion protein tagged with either GFP or mCherry to allow confirmation of intracellular localization (Fig. 4A). Co-expression of Parv-GFP-nesprin reduced ;82% the amplitude of the Ca 21 transient induced by depolarization in the perinuclear compartment, but did not affect Ca 21 transients detected by the diffusely localized parent RCaMP1h sensor (Fig. 4B-E), demonstrating that nesprin-1a localized parvalbumin could only reduce [Ca 21 ] in that compartment. Importantly, expression of the mCherry-Parv-nesprin fusion protein suppressed 66% KCl-induced PKA activity detected by PN-AKAR4 (Fig.  4F,G), demonstrating that elevation of [Ca 21 ] within the perinuclear compartment is required for full activation of PKA in mAKAPa signalosomes.

Perinuclear Ca 21 is required for activity-dependent neurite extension
Given that elevated perinuclear [Ca 21 ] was required for activation of mAKAPa-bound PKA, that we previously showed to regulate axon outgrowth (Boczek et al., 2019); we then asked whether selective chelation of perinuclear Ca 21 would inhibit axon outgrowth. Hippocampal neurons were transfected with plasmids to co-express either Parv-GFP-nesprin or control GFP-nesprin with mCherry, that served as a whole cell marker (Fig. 5A). Measurement of the longest neurite showed that in the absence of KCl, axon length was similar for GFP-nesprin and Parv-GFP-  RCaMP1h and for PN-AKAR4 and PN-CaMP1h. Here and below, a cytosolic region of interest in the soma was measured for the non-localized AKAR4 and RcAMP1h sensors. Scale bar: 10 mm. See Figure 3 for quantification of average responses. E, Averaged trace for PN-AKAR4 response to increasing KCl concentration (10, 30, and 40 mM). Solid line and shaded area indicate mean and SEM, respectively; n = 9 from four independent experiments. F, PN-AKAR4 amplitude to different KCl concentrations. Black bars indicate mean values. Datasets were compared by one-way ANOVA and Tukey's post hoc testing; **p 0.01, ***p 0.001. nesprin expressing neurons. KCl stimulation for 2 d induced a 15% increase in axon extension for control GFPnesprin neurons (Fig. 5B,C). In contrast, KCl-stimulation induced no increase in axon extension for neurons expressing Parv-GFP-nesprin, demonstrating that perinuclear Ca 21 signaling is necessary for activity-enhanced neurite extension.

Discussion
Using live cell imaging of primary hippocampal neurons, mAKAPa-bound PKA at the nuclear envelope is shown here to be activated by KCl-mediated depolarization via a mechanism requiring L-type Ca 21 channel activity and local increases in [Ca 21 ], promoting neurite extension (Fig. 6). This study extends prior observations regarding mAKAPa signalosomes and neurite extension, including (1) that mAKAPa expression and perinuclear localization is important for neurite extension in vitro (Wang et al., 2015;Boczek et al., 2019); (2) that elevated cAMP at mAKAPa signalosomes is sufficient and necessary to induce hippocampal neurite extension in vitro (Boczek et al., 2019); and (3) that displacement of the mAKAPabound PDE PDE4D3 results in elevated perinuclear cAMP levels and increased neurite extension (Boczek et al., 2019). Surprisingly, KCl-mediated membrane depolarization, which induced neurite outgrowth, increased PKA activity detected with the localized PN-AKAR4 sensor, but . Perinuclear Ca 21 is required for cAMP elevation at the nuclear envelope. A, Cyprinus carpio b -parvalbumin-nesprin-1a fusion proteins. Nesprin-1a contains five spectrin repeats (SRs) and a transmembrane KASH domain that localizes the protein to the nuclear envelope via binding to SUN domain proteins (Pare et al., 2005a). B, C, Hippocampal neurons expressing PN-RCaMP1h and either Parv-GFP-nesprin (Parv; n = 15) or control GFP-nesprin (Ctrl; n = 15) were stimulated with 40 mM KCl. D, E, Neurons expressing RCaMP1h and either Parv-GFP-nesprin (Parv; n = 26) or control GFP-nesprin (Ctrl; n = 23) were stimulated with 40 mM KCl. F, G, Neurons expressing PN-AKAR4 and either mCherry-Parv-nesprin (Parv; n = 26) or control mCherry-nesprin (Ctrl; n = 18) were stimulated with 40 mM KCl. Traces show mean 6 SEM and are normalized to initial baseline values (R 0 or I 0 ); black bars in C, E, F indicate mean values. Datasets compared by Mann-Whitney U test; ***p 0.001. not the diffusely expressed parent AKAR4 sensor, despite using a strong KCl stimulus. This was in contrast to prior findings that forskolin activated PKA detected with both sensors (Boczek et al., 2019). Further, relatively high levels (40 mM) of KCl was required for perinuclear PKA activation, consistent with previous findings that mAKAPadependent perinuclear signaling is linked to signaling in stressed, but not healthy, neurons (Wang et al., 2015). Neuronal activity, modeled in vitro by KCl-mediated depolarization, is a major determinant of CNS neurite extension and neuronal survival, and, moreover, induces these processes via cAMP/PKA-dependent mechanisms (Lipton, 1986;Shen et al., 1999;Goldberg et al., 2002;Corredor et al., 2012). Given additional prior findings regarding the role of mAKAPa signalosomes in retinal ganglion cell survival (Wang et al., 2015;Boczek et al., 2019), we suggest that the data shown herein support a model in which the perinuclear, mAKAPa cAMP compartment is a major node in the intracellular signaling network controlling both axon extension and neuroprotection.
Imaging of neurons expressing a nuclear envelope-localized parvalbumin fusion protein suggests that Ca 21 influx induced by KCl depolarization must include elevation of Ca 21 at the nuclear envelope in order for mAKAPa-bound PKA to be fully activated. In addition, depletion of Ca 21 at the nuclear envelope inhibited KCl-stimulated axon elongation, demonstrating the functional consequence of Ca 21 signaling within the perinuclear compartment. Notably, activation of L-type Ca 21 channels appears critical for this  process, consistent with the previously recognized role of these channels in regulating neuronal gene expression (Wild and Dell'Acqua, 2018). L-type Ca 21 channels have also been linked to hippocampal survival signaling in response to iron toxicity (Bostanci and Bagirici, 2013) and have been contrasted with NMDA-mediated Ca 21 entry and induction of cell death in hippocampal neurons (Stanika et al., 2012), although neither of these studies examined Ca 21 or cAMP signaling at the perinuclear region. Together with our data demonstrating dependence on L-type Ca 21 channels for perinuclear Ca 21 and cAMP signaling, and previous observations identifying the importance of this compartment for neuronal survival and axon growth (Boczek et al., 2019), these examples support a model in which specific Ca 21 signaling pathways converge on mAKAPa at the nuclear envelope to support survival and growth signaling. We cannot exclude that L-type Ca 21 channels at sites remote from the nuclear envelope regulate mAKAPa-bound PKA, including dendritic channels important for excitation-transcription coupling (Oliveria et al., 2007). However, our data are consistent with a model in which local influx through L-type Ca 21 channels that are near the nucleus confer compartment-specific activation. L-type channels are enriched on the soma of hippocampal neurons (Hell et al., 1993). The Dell'Acqua laboratory has elegantly demonstrated that somatic L-type channels induce nuclear factor of activated Tcells type 3 (NFATc3) transcription factor nuclear translocation via activation of the phosphatase calcineurin (Wild et al., 2019). While NFATc3 translocation in neurons does not appear to be dependent on ryanodine receptors that confer Ca 21 -induced Ca 21 release from intracellular stores (Wild et al., 2019), mAKAPb in striated myocytes has been shown to bind ryanodine receptors, and L-type channels can induce ryanodine receptor opening and release of stored Ca 21 (Marx et al., 2000;Kapiloff et al., 2001;Ruehr et al., 2003). Whether ryanodine receptors that have been detected at neuronal nuclear envelope and can regulate nuclear Ca 21 participate in elevating perinuclear Ca 21 fluxes at mAKAPa signalosomes will be subject of future studies (Walton et al., 1991;Kumar et al., 2008).
Our findings imply that a Ca 21 -dependent AC is responsible for local synthesis of cAMP in the mAKAPa compartment (Fig. 6). It is formally possible that Ca 21 -activates perinuclear cAMP via inhibition of a local cAMP PDE, albeit the PDE that regulates the mAKAPa compartment PDE4D3 is not known to be inhibited by Ca 21 signaling. Instead, the binding of an AC by mAKAPa could confer this local regulation. AC1, AC3, and AC8 are activated by Ca 21 /calmodulin, and AC10 (soluble AC) by Ca 21 and bicarbonate . mAKAP has been shown to bind AC5 and AC2, but not AC1 and AC6 when co-expressed in heterologous cells (Kapiloff et al., 2009). Other ACs were not tested for mAKAP binding. mAKAP residues 245-340 binds directly the conserved C1 and C2 catalytic domains of AC5, such that the specificity in mAKAP-AC binding presumably depends on AC sequences not conserved among isoforms. While activating ACs 1-8, forskolin does not activate AC9 and AC10 ). As both forskolin and KCl stimulate PN-AKAR4 in hippocampal neurons (as shown here and in Boczek et al., 2019), one might predict that AC3 or AC8 (but not AC1) is responsible for mAKAPa-associated PKA activity. However, forskolin should broadly activate transmembrane ACs in neurons, potentially resulting in a non-specific, non-physiologic diffuse cAMP activation, including in the mAKAPa compartment. AC10 can promote retinal ganglion cell neurite extension and survival in vitro (Corredor et al., 2012), and thus AC10 could participate in mAKAPa signalosomes. On the other hand, AC1/AC8 double knock-out did not affect retinal ganglion cell axon growth, but did inhibit the forskolin-potentiated survival of these neurons in vitro (Corredor et al., 2012). In addition, other ACs are regulated indirectly by Ca 21 -dependent protein kinases and phosphatases. Provocatively, while AC1/8 double knockout reduced KCl-dependent cyclase activity in hippocampal neurons ;60%, KCl could also activate AC in these cells via activation of calcineurin (Chan et al., 2005). As mAKAPb binds active calcineurin promoting the dephosphorylation of NFATc and MEF2 transcription factors (Dodge-Kafka et al., 2019), it is possible that KCl and Ca 21 activates AC in the mAKAPa compartment via a calcineurin-dependent pathway. The identification of the AC(s) critical for perinuclear cAMP-dependent neurite extension and neuroprotection will require future studies involving specific interference with the expression (RNAi) of individual cyclases and PN-AKAR4 imaging. While it remains to be established in neurons, given the prominent role of mAKAPb signalosomes in the control of stress-regulated cardiac myocyte gene expression (Dodge-Kafka et al., 2019), cAMP-dependent signaling at mAKAPa signalosomes presumably regulates neuronal gene expression controlling neurite extension. Future studies will be directed at the discovery of mechanisms by which activity-dependent cAMP signaling at mAKAPa signalosomes promote hippocampal neuron neurite outgrowth in vitro. Additionally, future studies should explore whether organization of signaling downstream of physiologic (e.g., synaptic) signaling is involved in other phenotypes including homeostatic regulation of activity. Meanwhile, the identification of a cAMP compartment that can promote axon growth and neuroprotection suggests that further study of this compartment is warranted in terms of both basic mechanism and potential translational relevance.