MicroRNA-134 activity in somatostatin interneurons
regulates H-Ras localization by repressing the
palmitoylation enzyme, DHHC9
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Chai, S., X. A. Cambronne, S. W. Eichhorn, and R. H. Goodman.
“MicroRNA-134 Activity in Somatostatin Interneurons Regulates
H-Ras Localization by Repressing the Palmitoylation Enzyme,
DHHC9.” Proceedings of the National Academy of Sciences 110,
no. 44 (October 14, 2013): 17898–17903.
As Published
http://dx.doi.org/10.1073/pnas.1317528110
Publisher
National Academy of Sciences (U.S.)
Version
Final published version
Accessed
Thu May 26 00:26:21 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/89106
Terms of Use
Article is available under a Creative Commons license; see
publisher’s site for details.
Detailed Terms
MicroRNA-134 activity in somatostatin interneurons
regulates H-Ras localization by repressing the
palmitoylation enzyme, DHHC9
Sunghee Chai1,2, Xiaolu A. Cambronne1, Stephen W. Eichhorn3, and Richard H. Goodman2
Vollum Institute, Oregon Health and Science University, Portland, OR 97239
Contributed by Richard H. Goodman, September 18, 2013 (sent for review July 30, 2013)
MicroRNA-134 (miR-134) serves as a widely accepted model for
microRNA function in synaptic plasticity. In this model, synaptic
activity stimulates miR-134 expression, which then regulates dendrite growth and spine formation. By using a ratiometric microRNA
sensor, we found, unexpectedly, that miR-134 activity in cortical
neurons was restricted to interneurons. Using an assay designed
to trap microRNA–mRNA complexes, we determined that miR-134
interacted directly with the mRNA encoding the palmitoylation
enzyme, DHHC9. This enzyme is known to palmitoylate H-Ras,
a modification required for proper membrane trafficking. Treatment with bicuculline, a GABAA receptor antagonist, decreased
DHHC9 expression in somatostatin-positive interneurons and
membrane localization of an H-Ras reporter in a manner that
depended on miR-134. Thus, although miR-134 has been proposed
to affect all types of neurons, we showed that functionally active
miR-134 is produced in only a selected population of neurons
where it influences the expression of targets, such as DHHC9, that
regulate membrane targeting of critical signaling molecules.
palmitoyltransferase
| RISC-trap
C
onsiderable progress has been made in understanding the
contributions of microRNAs to neuronal development, but
microRNAs are also found in mature neurons, where their localization in dendrites, and possibly axons, has been proposed to
contribute to synaptic plasticity (1–5). This is an appealing concept because local regulation of protein translation is essential
for plasticity and microRNAs can regulate dendritic growth,
spine formation, and growth cone guidance, properties generally
believed to underlie plasticity. Activity-regulated microRNAs are
particularly well suited to participate in these mechanisms. One
of the best-studied activity-dependent microRNAs in brain is
microRNA-134 (miR-134), which has been shown to regulate
dendritic spine morphogenesis through effects on LIM domain
kinase 1 (Limk1) (1) and dendrite growth through the translational
repressor, pumilio 2 (Pum2) (5). Mutations in LimK1 have been
linked to a human mental retardation syndrome (6), and Pum2deficient mice showed behavioral abnormalities (7). Recent studies
have demonstrated involvement of miR-134 in long-term potentiation (LTP) via regulation of sirtuin 1 (SIRT 1) and in epileptogenesis (8, 9). Like most studies involving brain microRNAs,
these reports make the implicit assumption that miR-134 is
expressed uniformly in excitatory neurons. It is well known,
however, that brain neurons are highly heterogeneous, and in
many regions, excitatory (largely glutamatergic) and inhibitory
(largely GABAergic) interneurons form complex networks that
balance opposing actions (10). Although many microRNAs are
probably expressed in both excitatory and inhibitory neurons,
some are likely to be restricted to one or the other subclass.
Indeed, analysis of Argonaute (Ago)2 complexes in individual
neuronal subtypes within the mouse brain identified a significant number of microRNAs that were enriched in GABAergic
but not in glutamatergic neurons and vice versa, indicating that
microRNA expression patterns are specific for neurons that share
a common origin and neurotransmitter phenotype (11). Thus,
17898–17903 | PNAS | October 29, 2013 | vol. 110 | no. 44
characterizing the expression patterns of microRNAs within a
heterogeneous population of neurons requires approaches that
monitor microRNA function at the single-cell level.
In previous studies, we used a ratiometric microRNA sensor to
monitor another activity-regulated microRNA, miR-132 (12). This
allowed us to quantify miR-132 function in individual cells by
comparing the ratios of GFP (fused to a sequence containing the
microRNA binding site) to an internal control, DsRed, driven by
a shared promoter. Using a similar sensor designed to monitor
miR-134, we found that unlike miR-132, miR-134 was not active in
the majority of cortical neurons. Rather, activity was detected in
cells that also stained for neuropeptides somatostatin (SST) or
calretinin. This indicated that miR-134 activity was largely limited
to inhibitory GABAergic interneurons. The finding that functionally active miR-134 was produced in only a limited population
of neurons must be considered in analysis of potential mRNA
targets. Using the RNA-induced silencing complex (RISC)-trap
method, designed to characterize microRNA–mRNA interactions
biochemically (13), we identified the transcript encoding the palmitoylation enzyme, DHHC9, as a putative miR-134 target. Palmitoylation is a lipid modification critical for proper trafficking of
membrane proteins (14). In most instances, the relationship between a particular palmitoylation enzyme and its targets is poorly
defined, in that each enzyme has multiple targets and each target
can be palmitoylated by multiple enzymes. This does not appear to
Significance
Most brain regions comprise numerous neural cell types that are
distinct in function and molecular characteristics. We examined
microRNA-134 (miR-134) activity in cortical cultures using
a microRNA sensor and discovered that miR-134 was induced in
response to neuronal activity in somatostatin- and calretininexpressing interneurons but not in pyramidal neurons. We
identified the palmitoylation enzyme DHHC9 as a direct target
of miR-134 and showed that it contributed to the regulation of
H-Ras in somatostatin interneurons. H-Ras is a signaling molecule crucial for neuronal development and function. This study
uncovered an activity-dependent miR-134 regulatory pathway
in cortical interneurons that can potentially affect membrane
localization of multiple signaling molecules.
Author contributions: S.C., X.A.C., and R.H.G. designed research; S.C., X.A.C., and S.W.E.
performed research; S.C., X.A.C., and S.W.E. analyzed data; and S.C., X.A.C., and R.H.G.
wrote the paper.
Conflict of interest: X.A.C. was involved in the commercial licensing of the RISC-trap
technology. This did not influence the experimental design, data collection, or analysis
of the project data.
1
S.C. and X.A.C. contributed equally to this work.
2
To whom correspondence may be addressed. E-mail: goodmanr@ohsu.edu or chaisu@
ohsu.edu.
3
Present address: Whitehead Institute for Biomedical Research, Massachusetts Institute of
Technology, Cambridge, MA 02142.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1317528110/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1317528110
be the case for H-Ras, which appears to be palmitoylated predominantly by DHHC9 (15). This modification is required for
directing H-Ras to the neuronal membrane, where it mediates
morphological changes and may participate in LTP and memory
formation (16). Our studies indicate that in SST+ neurons,
DHHC9 is down-regulated by neuronal activity through a pathway
that involves miR-134. The consequent reduction of DHHC9
expression disrupted localization of H-Ras from plasma membrane microdomains. Thus, although previous studies have highlighted miR-134 as an activity-dependent microRNA localized in
the synaptodendritic compartment, we show that functionally active miR-134 is produced in only a selected population of neurons
where it influences the expression of targets, such as DHHC9,
which influence the membrane targeting of critical signaling
molecules such as H-Ras.
Results
Previous studies in hippocampal and cortical cultures showed elevated levels of the miR-134 precursor transcript (pre–miR-134)
following either treatment with brain-derived neurotrophin factor
(BDNF) or KCl-induced membrane depolarization (1, 5). Using
cultures of rat hippocampal neurons (DIV8), we compared the
BDNF-stimulated induction of miR-134 with that of miR-132,
another activity-dependent microRNA involved in the development of dendrites and dendritic spines (17, 18). Using
quantitative PCR (qPCR) and Taqman microRNA assays to
monitor levels of precursor and mature microRNAs, respectively,
we detected relatively small increases in pre–miR-134 and mature
miR-134 following BDNF treatment over a time course of 6 h
(Fig. 1A). Levels of pre–miR-134 peaked 2 h after BDNF treatment (1.6 ± 0.5-fold above basal levels) and returned to baseline
by 6 h. Correspondingly, mature miR-134 levels were not significantly elevated above basal level after 6 h. In contrast, pre–miR132 levels increased by 18-fold over the same time period. Similar
results were obtained after treatment with 16 mM KCl. We hypothesized that induction of miR-134 may be restricted to specific
cell types and therefore may not be apparent in assays using
heterogeneous populations of cells. To assess the induction of
mature miR-134 in individual cells, we generated a ratiometric
sensor specific for miR-134, following a strategy that we had used
previously for measurement of miR-132 (Fig. 1B). This approach
monitors miR-134 activity through its regulation of a GFP transcript containing three reiterated miR-134 microRNA recognition elements (MREs) in its 3′UTR. Comparing GFP levels to
that of an internally expressed DsRed control lacking these
MREs provides a ratiometric measurement for microRNA activity. To test this system, we first assessed the efficacy of our
miR-134 sensor in SH-SY5Y neuroblastoma cells using flow
cytometry to analyze the GFP/DsRed ratio in individual cells.
These studies revealed a dose-dependent response of the miR134 sensor to exogenous miR-134 (Fig. 1C). A control sensor
lacking MREs was completely unresponsive to added miR-134.
The effect of exogenous miR-134 was also apparent at the sensor
transcript level (Fig. 1D). To monitor miR-134 induction in individual cortical neurons, we analyzed activity of the miR-134
sensor in primary cultures characterized using specific molecular
markers. In particular, we examined cells that stained for the neuropeptide, somatostatin, which represent a well-characterized subtype of inhibitory interneuron. Cortical cultures were transfected with
either the miR-134 sensor or control at DIV7, stimulated at DIV8
with bicuculline (a GABAA receptor antagonist that blocks GABAA
receptors and thereby is predicted to increase neural excitability), and
examined at DIV9 using relative intensity measurements (Fiji software). The population of SST-expressing interneurons was identified
by immunostaining (Fig. 1E). The GFP/DsRed ratio for the control
sensor in these cells had a mean value of 1.09; the ratio for the miR134 sensor was 0.62, indicating a robust increase in miR-134 activity
Chai et al.
Fig. 1. Activity-dependent induction of miR-134 is limited to interneurons. (A) BDNF marginally induced expression of pre– and mature miR-134.
qPCR (pre–miR-134 and pre–miR-132) and Taqman (mature miR-134)
assays were performed using RNA isolated from DIV7 cortical cultures
treated with 40 ng/mL BDNF for the times indicated. Expression levels were
normalized to actin mRNA, and fold induction represents the increase over
the 0 time value (*P = 0.02, ***P < 0.0001). (B) Ratiometric microRNA sensors. Sensor-134MRE contained three copies of miR-134 MRE downstream
of AcGFP. (C) Flow cytometric analysis of microRNA sensors coexpressed
with various amounts of miR-134 mimic in SH-SY5Y cells. The GFP/DsRed
ratio of individual cells was plotted as a cumulative frequency distribution.
(D) Introduction of miR-134 decreased GFP mRNA levels of the miR-134
sensor in HEK293 cells, analyzed by qPCR and normalized to internal control,
DsRed. (E) Analysis of miR-134 expression in bicuculline-treated SST+ neurons. Bicuculline treatment reduced GFP intensity in neurons containing the
sensor-134MRE, indicating miR-134 induction. (F) Quantification of the GFP/
DsRed ratio in individual SST+ (Left) and pyramidal (SST-negative) (Right)
neurons after bicuculline treatment. Intensity of green and red fluorescence
from confocal images was analyzed by Fiji software. The average GFP/DsRed
ratio per cell is graphed. ***P < 0.0001, n.s. P = 0.385. (G) Induction of miR134 depends on bicuculline treatment in SST+ (Upper) and calretinin neurons
(Lower). The GFP/DsRed ratios in SST+ and calretinin neurons containing
the sensor-134MRE decreased after bicuculline treatment. ***P < 0.0001,
*P = 0.01, n.s. P > 0.05. Student t test; histograms show mean ± SEM. n.s.,
not significant.
(Fig. 1F). We next analyzed sensor activity in neighboring large
excitatory pyramidal neurons. In these cells, the difference between the control and miR-134 sensors was not significantly different (P = 0.385; Fig. 1F). In contrast, miR-132 activity measured
using a specific sensor was readily apparent in these pyramidal
neurons (Fig. S1). These results indicated that miR-134 is induced
in SST+ neurons but not in pyramidal neurons, likely explaining
the modest increase of miR-134 in preparations of heterogeneous
cortical cultures where SST-expressing cells constitute a relatively
small minority. To determine whether miR-134 levels were, in
fact, induced after stimulation, we repeated the assays in the
presence or absence of bicuculline. With the control sensor,
bicuculline treatment did not significantly alter the GFP/DsRed
ratio (Fig. 1G). However, the miR-134 sensor in SST+ neurons
responded robustly (Fig. 1G), demonstrating that miR-134 induction in SST+ neurons is activity-dependent. To determine
whether miR-134 activity was induced in other subtypes of interneurons, we examined calretinin-positive interneurons. SST- and
PNAS | October 29, 2013 | vol. 110 | no. 44 | 17899
CELL BIOLOGY
Activity-Dependent Induction of miR-134 in SST-Expressing Interneurons.
bicuculline treatment, and miR-134 activity was absent in untreated cells (Fig. 1G). These data suggested that miR-134 is
induced in at least two major subtypes of early developing
neocortical interneurons.
Palmitoylation Enzyme DHHC9 Is a Target of miR-134. We identified
DHHC9 as a putative miR-134 target by using the RISC-trap
assay that we had developed previously to capture miRNA–
mRNA interactions (13). This assay, performed in HEK293 cells,
uses an epitope-tagged dominant negative form of the RISC
component, GW182, to stabilize microRNA–mRNA complexes.
The endogenous DHHC9 transcript enriched by 4.5-fold in the
presence of miR-134, compared with another neuronal microRNA, miR-124 (Fig. 2A). To confirm that DHHC9 was a miR134 target, we fused the DHHC 3′UTR to a luciferase cDNA
and assayed luciferase activity in HEK293 cells and primary
hippocampal cultures. Luciferase activity was significantly downregulated by exogenous miR-134 in both types of cells (Fig. 2B).
Within the DHHC9 3′UTR, we identified three putative miR134 MREs with perfect seed matches (Fig. 2C, designated M1,
M2, and M3). Mutation of M1 and M3 individually partially
stabilized a FLAG-DHHC9-3′UTR reporter; mutation of both
sites rendered the reporter completely refractory to miR-134
regulation, compared with a scrambled microRNA and with the
DHHC9 reporter lacking its 3′UTR (Fig. 2D).
We established that DHHC9 was expressed in SST+ interneurons (Fig. S2). To determine whether DHHC9 protein was
regulated by endogenous miR-134, we generated a ratiometric
reporter similar to the microRNA sensors described above but
incorporating the DHHC9 3′UTR (Fig. 2E). This reporter was
expressed in cortical cultures at DIV7, and cells were treated
with bicuculline or control for 24 h and examined in both SST+
interneurons and pyramidal neurons on DIV11 using ratiometric
analyses of fluorescence intensity. In SST+ cells, bicuculline
significantly down-regulated GFP expression (Fig. 2E). In contrast, pyramidal neurons exhibited no difference in the mean
GFP/DsRed ratio following bicuculline treatment (Fig. 2E).
Fig. 2. miR-134 regulates DHHC9. (A) Identification of DHHC9 as a direct
miR-134 target using RISC-trap. FLAG-tagged dominant-negative GW182 was
expressed with miR-134 or miR-124 mimic in HEK293 cells. RNA–RISC protein
complexes were immunoprecipitated using an anti-FLAG antibody, and the
associated RNAs were analyzed by qPCR. Relative DHHC9 mRNA enrichment
reflected the increase compared with complexes containing miR-124 (*P =
0.028). (B) miR-134 decreased expression of reporters containing the DHHC9
3′UTR. Luciferase reporters containing the DHHC9 3′UTR or vector control
were expressed with 10 nM miR-134 mimic in HEK293 cells and hippocampal
neurons (HN). Renilla reporter activity was measured 24 h posttransfection
and normalized to control firefly luciferase activity (***P < 0.0001). (C) Putative MREs in the DHHC9 3′UTR (M1, M2, and M3) are indicated with
underlined seed sequences for miR-134. Three nucleotides in each of the
putative MREs were mutated (underlined in red). (D) Evaluation of putative
MREs. FLAG-tagged reporter contained the full-length coding region and 3′
UTR of DHHC9. FLAG-DHHC9 reporters with wild-type or mutant MREs were
expressed in HEK293 cells with miR-134 mimic or scrambled microRNA control. Mutations in M1 and M3 individually or in combination abolished miR134 regulation, as shown in Western blots and compared with the DHHC9
reporter lacking its 3′UTR. (E) DHHC9 3′UTR reporter was down-regulated
upon bicuculline treatment in SST+ neurons. (Upper) The ratiometric DHHC9
sensor contains the full-length DHHC9 3′UTR fused to AcGFP. DsRed is an
internal control. (Lower) Cortical neurons transfected with DHHC9-sensor
were treated or untreated with 20 μM bicuculline. ***P < 0.0001, n.s. P =
0.62. Student t test; histograms show mean ± SEM. n.s., not significant.
calretinin-positive neurons have been identified as the main interneuronal population postnatally (P0–P14) in rat neocortex,
whereas other major interneuronal subtypes such as parvalbumin
(PV)- and vasointestinal peptide-positive neurons emerge later
(19). The sensor assays in cortical culture showed that similar to
SST+ neurons, calretinin neurons expressed miR-134 upon
17900 | www.pnas.org/cgi/doi/10.1073/pnas.1317528110
H-Ras Palmitoylation Is Critical for Its Subcellular Localization. We
next asked whether the regulation of DHHC9 in SST+ neurons
by miR-134 affected membrane trafficking of the DHHC9 palmitoylation target, H-Ras. Most studies examining H-Ras localization have used reporter constructs, such as GFP fused to
the H-Ras coding sequence. To minimize confounding effects
from ectopic H-Ras expression (20), we generated a GFP fusion
that contained only the carboxyl-terminal H-Ras trafficking domain. This domain (designated HRasC24) has been shown to be
sufficient to localize H-Ras to cell membranes and includes two
cysteine residues (C181 and C184) that, when palmitoylated,
promote H-Ras trafficking (Fig. 3A) (21–24). We first examined
the subcellular localization of wild-type GFP-HRasC24 in SST+
interneurons, compared with a mutant isoform that could not be
palmitoylated (C181S, C184S). Confocal microscopy revealed a
distinct punctate pattern of wild-type GFP-HRasC24 throughout
the soma and processes of SST+ neurons (Fig. 3B). The distinct
microdomain localization of GFP-HRasC24 is indistinguishable
from the expression pattern of the GFP-tagged full-length H-Ras
(GFP-HRas) (Fig. S3). In contrast, the palmitoylation-defective
mutant (C181S, C184S) exhibited a diffuse distribution of green
fluorescence throughout the cell body and processes, consistent
with the nonspecific targeting of mutated GFP-HRasC24 due to the
absence of palmitoylation (24). To quantify the difference in GFPHRasC24 localization, we used linescan measurements of GFP intensity, captured across 30 μm of continuous dendritic length for
each SST+ neuron (Fig. 3C). The high-intensity peaks (max) represented the localization of GFP-HRasC24 in punta, and the lowerintensity values represented diffused GFP (min). These measurements showed a significantly different profile of GFP-HRasC24
Chai et al.
was comparable to that seen after expression of miR-134. Together, our data suggest that miR-134 regulates DHHC9dependent palmitoylation and is important for H-Ras localization in SST+ interneurons.
Bicuculline Treatment Changes H-Ras Localization in an miR-134–
Dependent Manner in SST+ Neurons. We next showed that bicu-
culline, like exogenous miR-134, altered localization of GFPHRasC24 and GFP-HRas full-length protein in SST+ neurons
(Fig. 5 A and B and Fig. S3). To determine whether the bicuculline effect depended upon miR-134, we examined GFPHRasC24 localization in cells cotransfected with a miR-134 locked
nucleic acid (LNA) inhibitor. The inhibitor LNA-miR-134, but
not a scrambled control, completely blocked the bicuculline effect
(Fig. 5 C–F). These data suggest that miR-134 regulates DHHC9dependent palmitoylation and is important for H-Ras localization
in SST+ neurons.
CELL BIOLOGY
Fig. 3. Palmitoylation is critical for localizing H-Ras in SST+ neurons. (A) The
C-terminal membrane-targeting domain of H-Ras contains two palmitoylation sites (designated Pal; C181 and C184) and a farnesylation motif (designated Far; CAAX). GFP fusions to the wild type and the palmitoylation
mutant (C181S, C184S) are illustrated. (B) Subcellular localization of GFPHRasC24 in SST+ neurons. Dendrites of SST+ neurons expressing either the
wild-type or the palmitoylation defective H-Ras membrane-targeting domain are shown; localization of the wild-type H-Ras to microdomains was
diminished by mutating the two palmitoylation sites. (C) Quantitative
analysis using linescan measurement of GFP intensity in 30 μm of continuous
dendritic length depicted in B. Each peak denotes microdomain localization;
relative distance reflects the span of the segment analyzed in μm. (D)
Quantification of linescan data was performed by measuring differences (Δ
intensity) between the maximum (max) and minimum (min) intensity of each
peak. The Δ intensity plotted represented the average of the five highest
peaks in each dendrite. The wild type and palmitoylation mutant had a significant difference in Δ intensity. ***P < 0.0001 (n =10 neurons). Student
t test, mean ± SEM. (Scale bars, 5 μm.)
(C181S, C184S) than the wild-type counterpart, with broader,
rolling peaks. Differences (Δ intensity) between adjacent max
and min value were averaged in each neuron (Fig. 3D). Comparison of the wild-type and mutant GFP-HRasC24 localization
patterns showed that localization of the mutant was significantly
more diffuse. This difference represents the maximal effect of
palmitoylation on trafficking of the reporter.
miR-134 Regulation of DHHC9 Regulates Trafficking of H-Ras in SST+
Neurons. To determine whether miR-134 affected GFP-HRasC24
localization, we examined the effects of scrambled control or miR134 mimics in dissociated cortical cultures stained immunohistochemically for SST expression. The distinct punctate pattern of
GFP-HRasC24 was retained in the presence of control microRNA,
whereas expression of miR-134 caused a diffused pattern of GFP
fluorescence in the SST+ cells (Fig. 4 A and B). This effect was
somewhat less pronounced than with the C181S, C184S palmitoylation mutant, so we asked whether DHHC9 depletion resulted in
a similar mislocalization. To examine the contribution of DHHC9
to GFP-HRasC24 localization in SST+ cells, we generated a plasmid
that encodes a short-hairpin RNA to DHHC9 (shDHHC9)
downstream of a tdTomato fluorescence marker. In HEK293
cells, shDHHC9 was capable of depleting coexpressed Flag-DHHC9
protein levels (Fig. 4 C and D) and endogenous levels of DHHC9
in Neuro2A cells (Fig. S4). We identified shDHHC9-expressing
cells by presence of the tdTomato fluorescence and analyzed
GFP-HRasC24 localization in the population that was also positive for SST. Introduction of shDHHC9 increased levels of diffused GFP fluorescence compared with control (Fig. 4 E and F).
Moreover, the effect of DHHC9 depletion on GFP-HRasC24
Chai et al.
Fig. 4. miR-134 and DHHC9 regulate H-Ras trafficking in SST+ neurons. (A
and B) H-Ras localization changes after exogenous miR-134 expression. (A)
Linescan measurement of GFP intensity in control- or miR-134–expressing
neurons. (B) Quantification of linescan data by measuring Δ intensity (***P =
0.0004, n =15 neurons). (C) DHHC9 shRNA blocked expression of epitopetagged DHHC9 in HEK293 cells. sh-control represents a nonspecific shRNA.
(D) Constructs of shDHHC9 and sh-control in plasmids expressing tdTomato
are shown. (E and F) DHHC9 regulates H-Ras localization in SST+ neurons.
Linescan measurement of GFP intensity in dendrites expressing GFP-HRasC24
along with sh-control or shDHHC9. (F) DHHC9 knockdown significantly reduced Δ intensity (**P = 0.0024, n = 11 neurons). GFP intensity was measured from 30 μm of continuous dendritic length (A, B, E, and F). Student t
test, mean ± SEM.
PNAS | October 29, 2013 | vol. 110 | no. 44 | 17901
Fig. 5. Bicuculline disrupts H-Ras microdomain localization via miR-134. (A
and B) Bicuculline treatment significantly changed GFP-HRasC24 localization
in SST+ neurons (***P = 0.0002, n =17 neurons). (C and D) LNA-miR-134
blocked the bicuculline-induced change in H-Ras localization (P = 0.75,
n =15 neurons). Neurons were transfected with the plasmid expressing GFPHRasC24 along with 50 nM LNA-miR-134. (E and F) Bicuculline effect is
specific to miR-134. Scrambled control oligomer did not block bicucullineinduced H-Ras mislocalization (**P = 0.006, n =12 neurons). GFP intensity
was measured by linescan in 30 μm of continuous dendritic length, and
microdomain localization of H-Ras was measured by quantification of the Δ
intensity. Student t test, mean ± SEM. n.s., not significant.
Discussion
Several studies have suggested that microRNAs contribute to the
establishment and maintenance of synaptic plasticity in developing
and mature neurons (1, 3, 25, 26). MicroRNAs regulated by
neuronal activity are particularly suited for this role. The first example of an activity-regulated microRNA in mammalian brain
was miR-132, which regulates expression of p250GAP, a GTPaseactivating protein involved in remodeling the actin cytoskeleton
within dendritic spines (17, 18). miR-134 has also been proposed to
regulate dendritic spine morphology and synaptic plasticity (1, 8).
miR-134 provides a compelling model for these functions because
unlike miR-132, it has been localized to the synaptodendritic
compartment. Of note, miR-132 has been shown to increase the
size and density of dendritic spines (27, 28), whereas miR-134
appears to have the opposite effect (1, 9). Because of their
seemingly antagonistic actions, we set out to determine whether
17902 | www.pnas.org/cgi/doi/10.1073/pnas.1317528110
miR-134 and miR-132 were, in fact, expressed concurrently.
Earlier studies indicated that miR-132 was expressed generally
throughout the brain, including large pyramidal neurons (17, 18,
27, 29). Because the preponderance of neurons in the cortex are
excitatory, we hypothesized that miR-134, at a minimum, would
be expressed in this neuronal subtype. Using a ratiometric
sensor approach, we confirmed that miR-134 was induced by
bicuculline but found that the induction was restricted and
included SST+ interneurons.
In previous studies, the participation of miR-134 in regulating
dendritic spines was determined by examining a complex mixture
of neurons from hippocampus and cortex (1, 8, 9). Unlike pyramidal neurons, mature interneurons often lack spines, and when
present, they are few in number with an irregular distribution (30,
31). Consequently, regulation of spine morphology may be less
important in interneurons than it is in excitatory, glutamatergic
neurons. It remains possible, however, that miR-134 could be involved in the degeneration of dendritic spines in early stages of
interneuron development. Indeed, some of the SST+ interneurons
did have a few dendritic filopodia, possibly reflecting their immaturity (Fig. S5). Nonetheless, our findings do not support the idea
that miR-134 contributes directly to dendritic spine morphology in
pyramidal neurons because our sensor does not detect miR-134
activity in this population. Conceivably, the spine phenotype observed in excitatory neurons could result from miR-134 activity in
SST+ interneurons because these neurons innervate the distal
dendrites of principal neurons, where they are believed to contribute to the integration of synaptic inputs and synaptic plasticity
(32). In this study, we tested SST- and calretinin-positive neurons,
which are found at relatively early postnatal days, but our experimental model did not allow us to test the interneuron subtypes
that emerge in more mature brain. It is possible that miR-134 is
expressed in those other interneuron subtypes as well. The analysis
of Ago2 complexes from mouse cortex showed that microRNA
expression profiles are largely maintained among PV-, SST-, and
glutamate decarboxylase GAD2-expressing interneurons but are
strikingly different from glutamatergic neurons (11).
The idea that an activity-regulated microRNA could control
expression of a protein palmitoyltransferase has two implications.
First, it suggests that a microRNA could regulate a battery of
synaptic proteins concurrently by controlling their delivery to
synaptic membranes. This mode of regulation would not require
the individual mRNAs to share a common microRNA recognition site because a single protein palmitoyltransferase can
modify multiple protein targets. Second, this process, because it
is indirect, would not require the microRNA and synaptic target
to be colocalized in the same subcellular compartment. This
could allow microRNAs to regulate proteins at the growth cone,
for example, a region generally not thought to be a prominent
site of mRNA translation. Thus, modulation of protein palmitoyltransferases by activity-regulated microRNAs could provide
an avenue for influencing synaptic function that could be particularly relevant for mature neurons. This mode of regulation is
not entirely unprecedented—miR-138 was shown to regulate expression of APT1, a depalmitoylation enzyme, in quiescent neurons, activating Gα13 in the synapse and causing spine shrinkage (3,
26). In contrast to miR-134, however, miR-138 expression is repressed by neuronal activity (3, 26). We used H-Ras localization to
monitor DHHC9 activity in SST+ neurons. Both shRNA-mediated
reduction of DHHC9 and exogenous expression of miR-134
delocalized H-Ras from microdomains. We propose that the
effect of miR-134 on H-Ras localization in SST+ neurons occurs
through DHHC9 down-regulation. Exactly how changes in H-Ras
localization contribute to interneuron function is unknown. In
pyramidal neurons, H-Ras activity is increased in glutamatestimulated dendritic spines and axonal growth cones (20, 33).
Additionally, H-Ras contributes to presynaptic neurotransmitter
release via the ERK/Synapsin I pathway (34). Conceivably, effects
Chai et al.
Materials and Methods
Primary cortical and hippocampal neurons were prepared from E20 Sprague–
Dawley rats (Charles River) and maintained in Neurobasal medium supplemented with B-27 and GlutaMax (Life Technologies). Where indicated,
neurons were treated with 20 μM bicuculline (Tocris Bioscience) or 16 mM KCl.
All animal experiments were approved by the Institutional Animal Care and
Use Committee at Oregon Health and Science University.
Immunocytochemistry and Confocal Microscopy. Primary cortical neurons
grown on poly-L-Lysine–coated coverslips in 12-well plate were fixed at DIV
8–12 in 4% (wt/vol) paraformaldehyde/PBS and permeablized and blocked in
blocking buffer containing 2% (vol/vol) horse serum, 1% BSA, and 0.5%
Triton X-100. Primary antibodies, goat anti-somatostatin (1:100; Santa Cruz
D-20) and mouse anti-calretinin (1:1,000; Swant), and secondary antibodies,
Alexa Fluor-647 Donkey anti-goat IgG and Alexa Fluor-647 goat anti-mouse
IgG (Life Technologies), were used in blocking buffer for immunodetection
1. Schratt GM, et al. (2006) A brain-specific microRNA regulates dendritic spine development. Nature 439(7074):283–289.
2. Kye MJ, et al. (2007) Somatodendritic microRNAs identified by laser capture and
multiplex RT-PCR. RNA 13(8):1224–1234.
3. Siegel G, et al. (2009) A functional screen implicates microRNA-138-dependent regulation of the depalmitoylation enzyme APT1 in dendritic spine morphogenesis. Nat
Cell Biol 11(6):705–716.
4. Natera-Naranjo O, Aschrafi A, Gioio AE, Kaplan BB (2010) Identification and quantitative analyses of microRNAs located in the distal axons of sympathetic neurons. RNA
16(8):1516–1529.
5. Fiore R, et al. (2009) Mef2-mediated transcription of the miR379-410 cluster regulates
activity-dependent dendritogenesis by fine-tuning Pumilio2 protein levels. EMBO J
28(6):697–710.
6. Hoogenraad CC, Akhmanova A, Galjart N, De Zeeuw CI (2004) LIMK1 and CLIP-115:
Linking cytoskeletal defects to Williams syndrome. Bioessays 26(2):141–150.
7. Siemen H, Colas D, Heller HC, Brüstle O, Pera RA (2011) Pumilio-2 function in the
mouse nervous system. PLoS ONE 6(10):e25932.
8. Gao J, et al. (2010) A novel pathway regulates memory and plasticity via SIRT1 and
miR-134. Nature 466(7310):1105–1109.
9. Jimenez-Mateos EM, et al. (2012) Silencing microRNA-134 produces neuroprotective
and prolonged seizure-suppressive effects. Nat Med 18(7):1087–1094.
10. Huang ZJ, Di Cristo G, Ango F (2007) Development of GABA innervation in the cerebral and cerebellar cortices. Nat Rev Neurosci 8(9):673–686.
11. He M, et al. (2012) Cell-type-based analysis of microRNA profiles in the mouse brain.
Neuron 73(1):35–48.
12. Magill ST, et al. (2010) microRNA-132 regulates dendritic growth and arborization of
newborn neurons in the adult hippocampus. Proc Natl Acad Sci USA 107(47):
20382–20387.
13. Cambronne XA, Shen R, Auer PL, Goodman RH (2012) Capturing microRNA targets
using an RNA-induced silencing complex (RISC)-trap approach. Proc Natl Acad Sci USA
109(50):20473–20478.
14. Huang K, El-Husseini A (2005) Modulation of neuronal protein trafficking and function by palmitoylation. Curr Opin Neurobiol 15(5):527–535.
15. Swarthout JT, et al. (2005) DHHC9 and GCP16 constitute a human protein fatty
acyltransferase with specificity for H- and N-Ras. J Biol Chem 280(35):31141–31148.
16. Arendt T, et al. (2004) Neuronal activation of Ras regulates synaptic connectivity. Eur J
Neurosci 19(11):2953–2966.
17. Vo N, et al. (2005) A cAMP-response element binding protein-induced microRNA
regulates neuronal morphogenesis. Proc Natl Acad Sci USA 102(45):16426–16431.
18. Wayman GA, et al. (2008) An activity-regulated microRNA controls dendritic plasticity
by down-regulating p250GAP. Proc Natl Acad Sci USA 105(26):9093–9098.
Chai et al.
of SST+ and calretinin-expressing neurons. Confocal microscopy was performed using Olympus FV1000 (Advanced Light Microscopy Core facility,
Oregon Health and Science University).
Image Analysis. For measurement of GFP and DsRed intensity from microRNA
sensors, Z-stack images of neurons were taken with 40× objectives and
compressed into a single image. In each neuron, two to three dendrites were
analyzed for intensity of fluorescence using Fiji software. Statistical analysis
was performed using paired t test with two-tailed P value by GraphPad Prism
software. For H-Ras localization, Z-stack images of neurons were taken with
60× objectives with twofold magnification. Images were compressed into
a single image for analysis by LineScan tool using Metamorph software. In
SST+ neuron, primary dendrite projecting at least 30 μm from the soma was
chosen to measure the intensity of GFP. Using peaks detected by linescan, Δ
intensity was calculated by subtracting the basal level intensity (min) from
the maximum intensity (max) of each peak. Five highest peaks were chosen
to calculate the average Δ intensity of a dendrite, and 10–15 neurons were
chosen for analysis of one sample group. Statistical analysis was performed
using paired t test with two-tailed P value by GraphPad Prism software.
Detailed information on cell culture and transfection, molecular method
including DNA constructs and mutagenesis, qRT-PCR, Western blot analysis,
and RISC-trap assay is provided in SI Materials and Methods.
ACKNOWLEDGMENTS. We thank Drs. Gail Mandel, Gary Banker, Philip
Stork, Haining Zong, Brian Jenkins, Anthony P. Barnes, and Shane Tillo for
helpful discussions in this study. We also thank Dr. Stefanie Kaech Petrie and
Aurelie Snyder (Advanced Light Microscopy Core facility, Oregon Health and
Science University) for valuable advice in analysis of confocal images. This
work is supported by National Institutes of Health Grants NS097317 and
MH099099 (to R.H.G.), NS076094 (to X.A.C.), and P30 NS069305.
19. Gonchar Y, Wang Q, Burkhalter A (2007) Multiple distinct subtypes of GABAergic
neurons in mouse visual cortex identified by triple immunostaining. Front Neuroanat
1:3.
20. Fivaz M, Bandara S, Inoue T, Meyer T (2008) Robust neuronal symmetry breaking by
Ras-triggered local positive feedback. Curr Biol 18(1):44–50.
21. Hancock JF, Magee AI, Childs JE, Marshall CJ (1989) All ras proteins are polyisoprenylated but only some are palmitoylated. Cell 57(7):1167–1177.
22. Prior IA, Hancock JF (2001) Compartmentalization of Ras proteins. J Cell Sci 114(Pt 9):
1603–1608.
23. Greaves J, Chamberlain LH (2007) Palmitoylation-dependent protein sorting. J Cell
Biol 176(3):249–254.
24. Rocks O, et al. (2005) An acylation cycle regulates localization and activity of palmitoylated Ras isoforms. Science 307(5716):1746–1752.
25. Fineberg SK, Kosik KS, Davidson BL (2009) MicroRNAs potentiate neural development.
Neuron 64(3):303–309.
26. Banerjee S, Neveu P, Kosik KS (2009) A coordinated local translational control point at
the synapse involving relief from silencing and MOV10 degradation. Neuron 64(6):
871–884.
27. Impey S, et al. (2010) An activity-induced microRNA controls dendritic spine formation
by regulating Rac1-PAK signaling. Mol Cell Neurosci 43(1):146–156.
28. Hansen KF, Sakamoto K, Wayman GA, Impey S, Obrietan K (2010) Transgenic miR132
alters neuronal spine density and impairs novel object recognition memory. PLoS ONE
5(11):e15497.
29. Luikart BW, et al. (2011) miR-132 mediates the integration of newborn neurons into
the adult dentate gyrus. PLoS ONE 6(5):e19077.
30. Lee W-C, et al. (2006) Dynamic remodeling of dendritic arbors in GABAergic interneurons of adult visual cortex. PLoS Biol 4(2):e29.
31. Ascoli GA, et al.; Petilla Interneuron Nomenclature Group (2008) Petilla terminology:
Nomenclature of features of GABAergic interneurons of the cerebral cortex. Nat Rev
Neurosci 9(7):557–568.
32. Markram H, et al. (2004) Interneurons of the neocortical inhibitory system. Nat Rev
Neurosci 5(10):793–807.
33. Harvey CD, Yasuda R, Zhong H, Svoboda K (2008) The spread of Ras activity triggered
by activation of a single dendritic spine. Science 321(5885):136–140.
34. Kushner SA, et al. (2005) Modulation of presynaptic plasticity and learning by the
H-ras/extracellular signal-regulated kinase/synapsin I signaling pathway. J Neurosci
25(42):9721–9734.
35. Lugli G, Torvik VI, Larson J, Smalheiser NR (2008) Expression of microRNAs and
their precursors in synaptic fractions of adult mouse forebrain. J Neurochem
106(2):650–661.
PNAS | October 29, 2013 | vol. 110 | no. 44 | 17903
CELL BIOLOGY
of the miR-134/DHHC9/H-Ras pathway on axonal growth and
other presynaptic mechanisms in SST+ interneurons could influence dendritic spine formation in principal cells. Finally, previous studies of miR-134 have largely addressed activities within
the synaptodendritic compartment (1, 5, 8, 9). We observed miR134 activity in the soma, suggesting additional functions. Consistent with our findings, miR-134 was not enriched in the synaptosome fraction of cortical neurons, compared with a total
homogenate (35). Although some actions of miR-134 might
be localized to the soma, downstream effects could nonetheless involve synaptic regions through the effects of DHHC9 on
H-Ras and other protein targets.