Activity- and mTOR-Dependent Suppression of Kv1.1 Channel mRNA Translation in Dendrites

See allHide authors and affiliations

Science  06 Oct 2006:
Vol. 314, Issue 5796, pp. 144-148
DOI: 10.1126/science.1131693


Mammalian target of rapamycin (mTOR) is implicated in synaptic plasticity and local translation in dendrites. We found that the mTOR inhibitor, rapamycin, increased the Kv1.1 voltage-gated potassium channel protein in hippocampal neurons and promoted Kv1.1 surface expression on dendrites without altering its axonal expression. Moreover, endogenous Kv1.1 mRNA was detected in dendrites. Using Kv1.1 fused to the photoconvertible fluorescence protein Kaede as a reporter for local synthesis, we observed Kv1.1 synthesis in dendrites upon inhibition of mTOR or the N-methyl-d-aspartate (NMDA) glutamate receptor. Thus, synaptic excitation may cause local suppression of dendritic Kv1 channels by reducing their local synthesis.

Pathways involving the mTOR kinase and its upstream phosphoinositide 3-kinase (phosphatidylinositol 3-kinase, or PI3K) are important for neuronal signaling, including long-term potentiation (LTP) and long-term depression (LTD) (14). Negative regulators of this pathway, tumor suppressors linked to tuberous sclerosis complex (TSC) and Peutz-Jeghers syndrome (5), regulate local synthesis in synaptodendritic compartments (4, 68). Whether the mTOR pathway regulates ion channel synthesis and localization is an interesting open question. Although Kv1 channels are primarily localized to axons (9), they also contribute to the rapidly inactivating (A-type) and slowly inactivating (D-type) Kv channels in somatodendritic regions of neurons in the brain [(10) and references therein], thereby controlling local excitability of dendritic branches (11, 12) and restricting calcium spike generation to dendrites (13). Whereas the T1 tetramerization domain mediates Kv1 axonal targeting (14), it is unknown whether specific mechanisms are used for dendritic localization of Kv1 channels.

We found that rapamycin increased Kv1.1 (by 50 ± 14%, P = 0.03, n = 3) but not Kv1.4 in the CA1 dendritic field of rat hippocampal slices (Fig. 1A and fig. S1A). Rapamycin also increased Kv1.1 in cortical and hippocampal neuronal cultures (Fig. 1B and fig. S1B). Consistent with mTOR activation by PI3K and its effector Akt/protein kinase B thereby causing phosphorylation of cap-dependent translation initiation factors and the p70 S6 kinase (6, 15), rapamycin reduced phosphorylation of p70 S6 kinase but not Akt (Fig. 1B and fig. S1B). Whereas mTOR activity is generally associated with increased protein synthesis, it can also down-regulate certain proteins (2, 1618), similar to the down-regulation of Kv1.1 here.

Fig. 1.

Rapamycin inhibition of mTOR increased Kv1.1 protein in central neuronal dendrites. (A) Kv1.1 (black, upper panel; red, lower panel) and Kv1.4 (blue, lower panel) distribution in hippocampal slices stained with the DNA dye SYTOX (green) after treatment with dimethyl sulfoxide (DMSO; control) or rapamycin (200 nM) for 75 min in oxygenated artificial cerebrospinal fluid (ACSF) at 37°C. Scale bar, 100 μm. DG, dentate gyrus. (B) Immunoblot of phospho-Akt, phospho-S6 kinase, Kv1.1, Kv1.4, and tubulin of cultured cortical neurons [days in vitro (DIV) 30] with or without exposure to rapamycin. (C) Left: Cultured hippocampal neurons (DIV 23) treated with DMSO (control) or rapamycin (Rap; 200 nM) double-labeled for Kv1.1 (green) and MAP2 (red). Scale bar, 20 μm. Right: Rapamycin increased Kv1.1 level in dendritic puncta; **P < 0.005. (D) Left: Surface Kv1.1 of unpermeabilized hippocampal neurons (DIV 21) in DMSO (control) or rapamycin (200 nM) was labeled with antibody to an extracellular epitope of Kv1.1 (green), followed by permeabilization and MAP2 staining (red). Scale bar, 20 μm. The Kv1.1 puncta in permeabilized neurons (C) were “hotspots” that partially colocalized with mTOR granules (fig. S2B). Right: Rapamycin increased Kv1.1 surface expression in distal dendrites; *P < 0.05. See SOM for quantifications for (C) and (D).

In cultured hippocampal neurons expressing the microtubule-associated protein MAP2 in dendrites (Fig. 1C) and neural filament M in axons (fig. S1C), rapamycin increased Kv1.1 in punctate structures in dendrites but not axons (Fig. 1C). These Kv1.1-containing puncta partially (60 ± 5%) colocalized with mTOR-containing granules (fig. S2B). Moreover, antibody that recognizes an extracellular epitope of Kv1.1 (19) revealed high levels of Kv1.1 on the surface of proximal apical dendrites in neurons treated with rapamycin but not in control neurons (Fig. 1D).

To determine whether Kv1.1 could be synthesized in dendrites, we first localized endogenous Kv1.1 mRNA to dendrites by in situ hybridization (Fig. 2A and fig. S2A). Our microarray analysis revealed that, for Kv1.1 mRNA, the level in synaptosomes normalized to that in hippocampi was greater than those for the dendritically targeted MAP2 and ARC (activity-regulated cytoskeletal associated protein) mRNA and substantially higher than those for transcripts for the synaptic protein NSF (N-ethylmaleimide–sensitive factor) and nuclear proteins such as histones (table S1). Similarly, the normalized ratio for Kv1.1 mRNA determined by real-time polymerase chain reaction (PCR) was comparable to that for MAP2 mRNA (20) (Fig. 2B). Moreover, in hippocampal neurons infected with a Sindbis viral construct programmed to express fluorescent Kv1.1, the exogenous Kv1.1 mRNA including the 3′ untranslated region (3′UTR) was present in granules in dendrites up to 110 μm from the soma (Fig. 2C).

Fig. 2.

Kv1.1 mRNA in hippocampal neuronal dendrites. (A) Kv1.1 mRNA in MAP2-positive (red) dendrites of hippocampal neurons (DIV 28) revealed by in situ hybridization using antisense (top) and sense control (bottom) probe against 3′UTR of Kv1.1 (green); scale bar, 20 μm. Boxed dendrites in left panels are shown at high magnification on the right; scale bar, 5 μm. (B) Real-time PCR revealed that the ratio of synaptosomal mRNA and total hippocampal mRNA (synaptosome/total hippocampus) for Kv1.1 and MAP2, but not Kv1.4, was greater than that for NSF. ***P < 0.001; quantifications given in SOM. (C) Left: Dendritic localization of EGFP-Kv1.1 RNA (top), but not EGFP RNA, both with polyadenylation sequence, revealed by in situ hybridization (ISH) using the same antisense probe for EGFP (DIV 21). The boxed dendrites are shown at high magnification on the right. RNA (ISH), black in top panel, red in merged panel; EGFP, green; MAP2, blue. Scale bar, 20 μm. Right: Normalized ISH signal of EGFP-Kv1.1 mRNA (solid bars) and EGFP mRNA (open bars). Mean signal intensities were averaged every 10 μm along the dendrite (end point indicated on the x axis) and divided by hybridization signals within the first 10 μm from the soma to normalize for expression. Error bars represent SEM; *P < 0.05, **P < 0.01, ***P < 0.001 (Student's t test for comparison with EGFP control, n = 13 dendrites).

Next, we devised a method to distinguish Kv1.1 synthesized in dendrites from Kv1.1 synthesized in soma and then transported to dendrites. We fused Kaede, a photoconvertible fluorescent protein (21), to the N terminus of Kv1.1 (Kaede-Kv1.1), followed with the 3′UTR of Kv1.1 (fig. S3). Ultraviolet (UV) light induces cleavage of Kaede, converting its fluorescence from green into red. Newly synthesized Kaede-Kv1.1 could then be identified by its green chromophore. To look for newly synthesized Kaede-Kv1.1 in Sindbis virus–infected neurons treated with rapamycin or PI3K inhibitors, we imaged live neurons before and after UV-induced photoconversion of Kaede-Kv1.1 into red fluorescent protein and observed newly synthesized Kaede-Kv1.1 appearing as green fluorescence over 1 hour, before a second UV exposure photoconverted newly synthesized Kaede-Kv1.1—but not background fluorescence—into red fluorescence. We found that treating neurons with rapamycin, or with the PI3K inhibitors wortmannin or LY294002, caused a marked increase in newly synthesized Kaede-Kv1.1 in dendrites (Fig. 3A). This was due to local synthesis because there was little movement of Kaede-Kv1.1: After photoconversion of the Kaede fluorescence within a dendritic branch, there was no detectable green Kaede-Kv1.1 in that region over 50 min, whereas Kaede fused to only the soluble N-terminal domain of Kv1.1 would have moved to the photoconverted dendritic region within 5 min (fig. S3). In rare occasions we observed Kaede-Kv1.1 and EGFP-Kv1.1 (Kv1.1 fused to enhanced green fluorescent protein) moving with the slow transport rate of less than 15 μm/hour. In general, the fluorescently tagged Kv1.1 in dendrites was concentrated in stationary “hotspots,” as reported for locally translated proteins (8, 22).

Fig. 3.

Inhibition of PI3K or mTOR increased local translation of Kaede-Kv1.1. (A) Live imaging of neurons in ACSF containing DMSO (control), rapamycin (200 nM), LY294002 (50 μM), or wortmannin (50 nM) before, immediately after (0 min), and 60 min after the first UV exposure to photoconvert Kaede-Kv1.1 into red fluorescent protein, for appearance of newly synthesized green Kaede-Kv1.1, which was then turned into red by the second photoconversion. Left panels: Representative grayscale images of green fluorescence in neurons. Scale bar, 20 μm. Right panels: Dendrite in orange boxed region with arrow pointing to a single Kaede-Kv1.1 puncta. Scale bar, 2 μm. (B) Normalized newly synthesized Kaede-Kv1.1 (green fluorescence pixel intensity) in individual puncta more than 50 μm from the soma, over the course of 1 hour, was increased by rapamycin, LY, or wortmannin. *P < 0.05; quantifications given in SOM.

After photoconversion, the newly synthesized Kaede-Kv1.1 grew over 1 hour at pre-existing hotspots without detectable movements (Fig. 3A, arrows). The newly synthesized Kaede-Kv1.1 that appeared in dendritic puncta more than 50 μm from the soma was increased by a factor of 3 in the presence of rapamycin or the PI3K inhibitors LY294002 or wortmannin (Fig. 3B). The protein synthesis inhibitors anisomycin and cycloheximide reduced this rapamycin effect (Fig. 4, B and C), supporting the notion of Kv1.1 local synthesis in the dendrites.

Fig. 4.

NMDA receptor inhibition increases dendritic Kv1.1 local translation and reduces mTOR phosphorylation. (A) Upper panels: NMDA receptor antagonist AP5 (200 μM) but not metabotropic glutamate receptor antagonist MCPG specific for mGluR1/mGluR5 (50 μM) treatment of hippocampal neurons (DIV 21 to 30) for 75 min at 37°C increased Kv1.1 total protein, decreased P-mTOR, and increased the ratio of Kv1.1 and P-mTOR (determined for each sample). Lower panel: Immunoblot of Kv1.1 (top) and P-mTOR (bottom). (B) Live imaging of Kaede-Kv1.1 in neurons exposed to rapamycin (200 nM), rapamycin plus anisomycin (40 μM), or cycloheximide (CHX; 5.5 μg/ml) before, immediately after (0 min), and 60 min after photoconversion and after a second photoconversion. Left panels: Representative grayscale images of green fluorescence in neurons. Scale bar, 20 μm. Right panels: Magnified view of orange boxed region. Scale bar, 2 μm. (C) Protein synthesis–dependent stimulation by rapamycin and AP5 of the normalized newly synthesized Kaede-Kv1.1 (green fluorescence pixel intensity) that appeared over 1 hour in puncta greater than 50 μm from the soma. *P < 0.05, **P < 0.01; see SOM for quantifications for (A) and (C).

To explore how Kv1.1 local translation might be regulated by neuronal activity, we examined neuronal cultures with and without treatments with antagonists of the N-methyl-d-aspartate (NMDA) glutamate receptor and metabotropic glutamate receptor. The NMDA receptor antagonist AP5 significantly increased endogenous Kv1.1 but decreased mTOR phosphorylation, raising the ratio of Kv1.1 protein to phosphorylated mTOR (P-mTOR) protein by a factor of 11 (Fig. 4A), a process that required protein synthesis (Fig. 4, B and C). It thus appears that NMDA receptors activated as a result of synaptic activity could locally regulate Kv1.1 synthesis near the active synapse.

We note that Kv1.1 mRNA does not contain internal ribosome entry sites (IRES) implicated in mTOR regulation, and the 5′UTR of Kv1.1 was not required for Kaede-Kv1.1 local synthesis regulation. The 3′UTR of Kv1.1 mRNA has two putative AU-rich elements (AREs) similar to those in the ligatin mRNA under NMDA receptor regulation (23). ARE binding proteins could either stabilize mRNA for translation or promote mRNA degradation, and they compete with one another for mRNA binding (24). It remains to be determined whether NMDA receptor activation shifts Kv1.1 mRNA from a form readily accessible to translation to a different complex for silencing or degradation.

How might NMDA receptors activate PI3K and mTOR to suppress Kv1.1? LTP is associated with regional increase of PI3K activity (25), probably because of NMDA receptor–mediated Ca2+ influx that activates PI3K associated with AMPA glutamate receptors nearby (26), and causes Akt and p70 S6 kinase phosphorylation essential for LTP expression but not induction or maintenance (25). NMDA receptor activation of PI3K at the active synapses could suppress Kv1.1 synthesis locally in the dendrites and may enhance excitatory post-synaptic potential (EPSP) spike potentiation (12).

NMDA receptor–mediated suppression of Kv1.1 local synthesis is a positive feedback mechanism that could specifically potentiate active synapses by enhancing voltage-gated sodium and/or calcium channel activation during EPSP, thus facilitating EPSP summation and action potential generation (11, 27) (fig. S5). Hippocampal neuronal dendrites exhibit uneven distribution of Kv channels (11, 28), one-fifth of which is sensitive to the Kv1-specific α-dendrotoxin (11), to prevent action potential initiation in dendrites and dampen backpropagation (11, 13). With the intriguing findings of clustered D-type Kv channels on CA1 dendrites (28) and localized increase of dendritic excitability and calcium influx within the dendritic branch with potentiated excitatory postsynaptic current (12), it is an interesting open question whether synaptic activities dynamically regulate the size and location of microdomains of Kv1 channels, thereby influencing integration and plasticity of synaptic inputs near recently active synapses to adjust the quality and capacity of information storage (27). It will also be important to elucidate homeostasis mechanisms of Kv1.1 local synthesis regulation for the resetting of local excitability.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Table S1


References and Notes

Stay Connected to Science

Navigate This Article