15394 The Journal of Neuroscience, November 12, 2014 34(46):15394 15401

Systems/Circuits

Visualizing the Engram: Learning Stabilizes Odor

Representations in the Olfactory Network
X Amin MD. Shakhawat,1* X Ali Gheidi,1* Qinlong Hou,1 Sandeep K. Dhillon,3 Diano F. Marrone,3 X Carolyn W. Harley,2
and Qi Yuan1
1

Division of Biomedical Sciences, Faculty of Medicine, and 2Department of Psychology, Faculty of Science, Memorial University, St. Johns, Newfoundland
A1B 3V6, Canada, and 3Department of Psychology, Faculty of Science, Wilfrid Laurier University, Waterloo, Ontario N2L 3C5, Canada

The nature of memory is a central issue in neuroscience. How does our representation of the world change with learning and experience?
Here we use the transcription of Arc mRNA, which permits probing the neural representations of temporally separated events, to address
this in a well characterized odor learning model. Rat pups readily associate odor with maternal care. In pups, the lateralized olfactory
networks are independent, permitting separate training and within-subject control. We use multiday training to create an enduring
memory of peppermint odor. Training stabilized rewarded, but not nonrewarded, odor representations in both mitral cells and associated granule cells of the olfactory bulb and in the pyramidal cells of the anterior piriform cortex. An enlarged core of stable, likely highly
active neurons represent rewarded odor at both stages of the olfactory network.
Odor representations in anterior piriform cortex were sparser than typical in adult rat and did not enlarge with learning. This sparser
representation of odor is congruent with the maturation of lateral olfactory tract input in rat pups. Cortical representations elsewhere
have been shown to be highly variable in electrophysiological experiments, suggesting brains operate normally using dynamic and
network-modulated representations. The olfactory cortical representations here are consistent with the generalized associative model of
sparse variable cortical representation, as normal responses to repeated odors were highly variable (70% of the cells change as indexed
by Arc). Learning and memory modified rewarded odor ensembles to increase stability in a core representational component.
Key words: Arc; olfactory bulb; olfactory learning; piriform cortex

Introduction
The rat pup odor preference learning model is highly attractive as
a tractable model of mammalian associative learning. The rodent
pup readily acquires preferences for odors paired with maternal
care signals to support maternal recognition (Logan et al., 2012).
The conditioned stimulus in this associative model is typically a
novel odor, whereas the unconditioned stimulus is provided by
norepinephrine (NE) release from the locus ceruleus acting
through an ensemble of noradrenergic receptors, the best studied
of which is the -adrenoceptor (Yuan et al., 2014). This NE release can be induced by tactile stimulation with a brush to mimic
maternal care (Rangel and Leon, 1995). A single trial in which
pups on peppermint-scented bedding are stimulated creates a
preference for peppermint lasting 24 h, whereas multiple trials

Received Aug. 14, 2014; revised Sept. 27, 2014; accepted Oct. 7, 2014.
Author contributions: D.F.M., C.W.H., and Q.Y. designed research; A.MD.S., A.G., Q.H., S.K.D., D.F.M., and Q.Y.
performed research; A.MD.S., A.G., and Q.Y. analyzed data; A.MD.S., A.G., C.W.H., and Q.Y. wrote the paper.
This work was supported by a Canadian Institutes of Health Research operating grant (MOP-102624) to Q.Y. and
a Natural Sciences and Engineering Research Council discovery grant (06609-2014) to D.F.M. We thank Dr. Jules Dore
for assisting in the initial establishment of Arc catFISH used in this study.
*A.MD.S. and A.G. contributed equally to this work.
The authors declare no competing financial interests.
Correspondence should be addressed to Dr. Qi Yuan, Division of Biomedical Sciences, Faculty of Medicine, Memorial University, St. Johns, NL A1B 3V6, Canada. E-mail: [email protected].
DOI:10.1523/JNEUROSCI.3396-14.2014
Copyright 2014 the authors 0270-6474/14/3415394-08$15.00/0

spaced over days creates more enduring memories (Fontaine et

al., 2013).
Cellular events critical for learning have been identified in
both the olfactory bulb (OB) and anterior piriform cortex (aPC).
Mechanisms for learning include activation of NMDA receptors
(NMDARs; Lethbridge et al., 2012; Morrison et al., 2013), L-type
calcium channels (Jerome et al., 2012), metabotropic glutamatergic receptors (Rumsey et al., 2001), adrenergic receptors (Sullivan
et al., 2000; Harley et al., 2006; Shakhawat et al., 2012; Morrison et
al., 2013), and disinhibition (Lethbridge et al., 2012). Intracellular changes critical for learning in the OB include a temporally specific cAMP transient (Cui et al., 2007), activation of
protein kinase A (Grimes et al., 2012), phosphorylation of
CREB (McLean et al., 1999), and an insertion of AMPA receptors (AMPARs; Cui et al., 2011).
Changes that relate to long-term memory expression are
fewer in number. Visualization methods have shown an increase
in intrinsic optical signaling (Yuan et al., 2002), an increase in
AMPARs at the glomerular level (Cui et al., 2011), and an increase in
network strength in the aPC (Fontaine et al., 2013). Electrophysiological methods have shown potentiation of the olfactory nerve to
mitral cell synapse in the OB (Yuan and Harley, 2012) and of the
lateral olfactory tract mitral cell output to an aPC pyramidal cell
synapse (Fontaine et al., 2013; Morrison et al., 2013).
Maintained increases in AMPAR strength, which have been
hard to demonstrate with memory in other systems, have been

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

J. Neurosci., November 12, 2014 34(46):15394 15401 15395

clearly seen in this model (Fontaine et al., 2013). The commissural connections are not mature in the 1-week-old rat pup, and
thus odor input is lateralized both in the OB and piriform cortex
(Kucharski et al., 1986; Kucharski and Hall, 1987; Fontaine et al.,
2013). Taking advantage of this within-animal control, AMPAR
changes congruent with memory duration were readily revealed
(Fontaine et al., 2013).
In the present study, cellular compartment analysis of temporal activity by fluorescence in situ hybridization (catFISH) of Arc
mRNA was used to identify odor ensemble representations in the
OB and aPC of rat pups that had undergone odor preference
training with one naris occluded. The outcomes support current
views of cortical representations in mammalian brain and suggest
stability of cell participation in representations is the signature
feature of learning and memory.

Materials and Methods

Animals. All experiments with animals were approved by the Animal
Care Committee of Memorial University of Newfoundland in compliance with the guidelines of the Canadian Council on Animal Care.
Sprague Dawley rat pups of both sexes were used in this study. Dams with
pups were housed in a vivarium that was temperature controlled and on
a 12 h light/dark cycle. The date of birth for the pups was designated
postnatal day 0 (P0).
Early odor preference training. The early odor preference training protocol with single naris occlusion has been established previously (Yuan
and Harley, 2012; Fontaine et al., 2013). Rat pups were assigned to one of
two conditions: odor paired with stroking (O/S ) or odor only (O/S ).
Four-day behavioral training was performed from P3 to P6. During
training, all pups received left naris occlusion for each session. Nose plugs
were constructed from polyethelyne-20 tubing (Yuan and Harley, 2012;
Fontaine et al., 2013). Pups were given a sterile 2% xylocaine gel application on the left naris 5 min before plug insertion. Pups were left to rest
for 5 min before subsequently being given either O/S or O/S training.
During training, pups were placed on peppermint-scented bedding (0.3
ml of peppermint for 500 ml volume of bedding). Pups in the O/S
group were simultaneously stroked with a paint brush (30 s stroking
interleaved with 30 s rest) for 10 min. Pups in the O/S were placed on
peppermint bedding for 10 min without being stroked. Nose plugs were
removed immediately after the training, and pups were returned to the
dams.
Tissue collection. On P7, pups were placed into covered plastic jars with
charcoal-filtered clean air flow for 1.5 h before being given two 5 min
odor deliveries separated by 20 min: either 2 peppermint or peppermint followed by vanillin or 2 vanillin (Fig. 1A). For odor delivery,
pups were moved to an adjacent covered jar with peppermint or vanillin
bedding at the bottom (0.3 ml of odor extract mixed with 500 ml of
normal bedding) and then switched back to the clean-air jar in the 20 min
interval. A naive group was used initially to test odor input specificity.
Pups in this group were exposed to two different odors without prior training. For this latter experiment (Fig. 1), 1% peppermint or vanillin odor
diluted in mineral oil was delivered through the air-delivery system (Knosys
olfactometer) for the 5 min odor periods (Shakhawat et al., 2014).
After the second odor exposure, rats were decapitated, and their brains
were flash-frozen in 2-methyl-butane immersed in an ethanol/dry ice
slurry. Brains were preserved in a 80C freezer until being sectioned at
20 m in a cryostat set at 20C. Sections of right hemispheres of the
animals in the input specificity study and both hemispheres of pups from
all other groups were mounted onto 2% 3-aminopropyltriethoxysilanetreated slides (Snowcoat; Leica) using OCT compound (Tissue-Tek;
Sakura Fintek USA). Each block usually contained four to six brains from
a particular experiment so that these brains were processed together. Five
to six slides taken evenly through the rostral to caudal range of the OB
and the aPC were used for fluorescent in situ hybridization and stored at
20C.
Fluorescence in situ hybridization. The fluorescence in situ hybridization protocol used was established previously (Guzowski and Worley,

Figure 1. Arc mRNA visualization reveals odor input-specific activation of mitral cell ensembles in the OB. A, Schematic of tissue collection protocols in naive and trained rat pups (top) and
example images for Arc cells (bottom). Blue indicates nuclei staining by DAPI. Red indicates
Arc staining. White arrows indicate Arc staining in nuclei. Yellow arrows indicate Arc cytoplasm
staining. Scale bar, 10 m. B1, Example image of Arc expression in the OB of a naive rats
exposed to 2 peppermint. GL, Glomerular layer; MCL, mitral cell layer; GCL, granule cell layer.
Scale bar, 500 m. B2, Example images of dorsolateral OB Arc expression in a naive rat pup to
two odor episodes. White arrows indicate Arc double cells in the MCL. Scale bars, 20 m. B3,
OLRs of the cell ensembles of the two odor episodes. **p 0.01, PP, Peppermint; VA, vanillin.
2001; Shakhawat et al., 2014). Briefly, Arc full-length DNA plasmid was
digested using EcoRI (Invitrogen) and run against a DNA ladder to confirm yield and base pair accuracy (2.5 kb). Digoxegenin-labeled riboprobes were synthesized from the digested DNA template using a
Maxiscript transcription kit (Ambion). Arc antisense riboprobe yields
were confirmed using 1% agarose gel electrophoresis. Slides were
brought to room temperature, fixed with 4% paraformaldehyde, bathed
with acetic anhydride and methanol/acetone (Thermo Fisher Scientific),
and treated with prehybridization buffer followed by hybridization buffer (Sigma-Aldrich) and Arc riboprobe. Hybridization occurred overnight in a 56C oven. The next day, after a series of sodium citrate washes,
any remaining single-stranded RNA was cleaved using Rnase A (SigmaAldrich) at 37C. Endogenous peroxidases were quenched with H2O2,
and slides were blocked with 5% sheep serum (Sigma-Aldrich) and incubated with anti-digoxegenin horseradish peroxidase (Roche) for 2 h.
After a series of Tris-buffered saline washes, the Cy3 fluorescent marker
(PerkinElmer) was applied to visualize Arc mRNA, and nuclei were
counterstained with 4-6-diamidino-2-phenylindole (DAPI; 1:2000;

15396 J. Neurosci., November 12, 2014 34(46):15394 15401

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

Sigma-Aldrich). Finally, slides were covered

with Vectashield antifade medium (Vector
Laboratories) and sealed with clear nail polish.
Slides were kept at 4C before confocal microscopy scanning.
Confocal image acquisition. Using an FV1000
confocal microscope (Olympus), optical z-sections
were taken from both the OB and the aPC. Images of mitral cell layers were taken at 40 with
two standardized areas (0.06 mm 2 each) in
the dorsolateral quadrant and two areas in the
ventromedial quadrant of the OB (Fig. 2A).
Images of pyramidal cell layers (II/III) of the
aPC were taken at 20. Two standardizedsized areas (0.3 mm 2 each; one in lateral and
one in medial aPC; Fig. 4A) were scanned. The
z-stacks (1.0 m thickness) throughout each
section (20 m) of the OB and the aPC were
acquired from three to four slides spread
evenly over the rostral to caudal range. Photomultiplier tube assignments, confocal aperture
size, and contrast remained constant for each
slide. The average counts of the two areas were
used for final counts for the dorsolateral and
ventromedial OB and for the aPC.
Image analysis. Off-line image analysis was
performed using ImageJ software. The total
numbers of DAPI cells were assessed using the
ImageJ automatic cell-counting application for
the aPC and the manual counting option for
the OB. Foci, cytoplasmic, and double labeling
of Arc-positive (Arc ) cells were counted manually. Labeling of cells as foci, cytoplasmic, and
double was achieved by checking multiple optical
sections (20% midrange of the z-stack) that comprised each individual cell (Miyashita et al.,
2009). Counting was performed by an individual
blind to all experimental training conditions.
Statistics. OriginPro 9.0 software was used to
analyze all data sets. Data were reported as the
mean SEM. Two-sample paired t tests were
used for statistical comparisons for all experiments except for the input specificity experiment in Figure 1 and the comparison of
occluded hemispheres across groups, in which
the two-sample unpaired t test was used. Differences between groups were considered significant when p values were 0.05.

Results

Figure 2. Early odor preference learning stabilizes the mitral cell ensemble to the conditioned odor in the OB. A, Schematic of OB

The immediate-early gene Arc has been anatomy and Arc sampling regions (red rectangles). D, Dorsal; V, ventral; L, lateral; M, medial; ON, olfactory nerve; GL, glomerular
established as a marker to index plasticity- layer; MCL, mitral cell layer; GCL, granule cell layer. B1B3, O/S training leads to increased overlap of mitral cell ensembles in the
related neuronal activation in multiple dorsolateral olfactory bulb responding to 2 peppermint exposures. B1, Example images of the mitral cell layer in the occluded
bulbs from the same animal. B2, OLR of mitral cell ensembles responding to 2 peppermint exposures. B3,
brain areas, including the olfactory cortex and spared olfactory

(Guzowski et al., 2005; Shakhawat et al., Percentage of Arc cells over the total population indexed by DAPI staining. C1C3, O/S training does not change the OLR of mitral cell
training leads to increased overlap of mitral
2014). Although previous research using ensembles in the dorsolateral OB responding to 2 peppermint exposures. D1D3, O/S
cellensemblesintheventromedialOBrespondingto2peppermintexposures.E1E3,O/S trainingdoesnotchangetheOLRofmitral
Northern blots suggested Arc was not ex- cellensemblesintheventromedialOBrespondingto2peppermintexposures.F1F3,O/S trainingwithpeppermintdoesnotchange
pressed early in development in the fore- the OLR of mitral cell ensembles in the dorsolateral OB responding to 2 vanillin exposures. MC, Mitral cell; DL, dorsolateral; VM, ventraobrain (Lyford et al., 1995), the more medial; PP, peppermint; VA, vanillin. Arrows indicate double-stained Arc cells. Scale bars, 20 m. *p 0.05; **p 0.01.
sensitive in situ hybridization technique
readily reveals the presence of Arc mRNA
sion (Fig. 1A). In the present experiments, we were also able to use
in our study. Arc transcription first appears in the neuronal nuArc to examine granule cells, although it is not normally often excleus within 5 min of neuronal activity. Thirty minutes later,
pressed in inhibitory interneurons (Vazdarjanova et al., 2006; Mcinitial Arc mRNA has trans-located to the cytoplasm, and a secCurry et al., 2010) and did not occur here in the juxtaglomerular
ond event can initiate new transcription of nuclear Arc (Guneurons.
zowski et al., 2005). Therefore, Arc permits discrimination of two
Two sets of experiments were included in this study. First,
separate odor events through analysis of compartmentalized expresnaive rat pups were used to test whether Arc can serve as an

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

input-specific activity marker in the OB. Second, rat pups underwent either odor paired with stroking (O/S ) or odor-only (O/S )
training and were given 2 peppermint or vanillin before brain
extractions (Fig. 1A).
Odor input specificity in the OB indexed by Arc mRNA
To test the odor input specificity of Arc activation, naive pups
were exposed to two 5 min episodes of odor: either peppermint
on both occasions separated by a 25 min interval (Fig. 1A, top,
PP-PP) or peppermint followed by vanillin 25 min later (Fig. 1A,
top, PP-VA). Animals were killed immediately after the second
episode and processed for Arc catFISH. Cells that expressed Arc in
the cytoplasm only were active during the first odor episode (peppermint) whereas cells that expressed Arc only in the nuclei were
active during the second odor episode (peppermint or vanillin),
and cells expressing Arc in both the nuclei and cytoplasm were
activated by both odor episodes (see example cells in Fig. 1A,
bottom).
Peppermint activated both mitral cells and granule cells in the
OB, especially the dorsolateral and ventromedial regions that
were previously shown as hot spots for peppermint (Johnson
and Leon, 1996; Fig. 1B1). Arc cells in the mitral cell layer were
counted in the dorsolateral region of the OB. On average, novel
peppermint activated 7.5% of the cells in the mitral cell layer of
the dorsolateral OB, whereas novel vanillin activated 6.4%
of the cells in the same region. Comparing the overlap ratio
(OLR; the proportion of cells with double staining relative to the
total number of Arc cells) of the cell ensembles activated by two
odor events, we demonstrated that repeated peppermint exposure was associated with significantly greater overlap (32.43
1.64%, n 4) than peppermint followed by vanillin exposure
(18.73 2.79%, n 4, t 4.23, p 0.006; Fig. 1B2,B3). This
experiment suggests that Arc mRNA can be used as a marker for
input-specific representations of odors in the OB. The same odor
is more likely to initiate Arc transcription twice in the same cells.
Odor preference training leads to more stable odor
representation in the mitral cell layer of the OB
We next trained rat pups in a multiday (P3P6) peppermint
O/S conditioning with a single naris occluded. The ensembles
of neurons responding to peppermint in the OB after training
were assessed by Arc mRNA expressions induced by two peppermint episodes (Figs. 1A, 2, PP-PP). The trained OB was compared with the occluded side to achieve an intra-animal control.
We have shown that single naris occlusion during multiday
training leads to lateralized learning and synaptic changes that
are confined to the spared olfactory hemisphere (Yuan and
Harley, 2012; Fontaine et al., 2013). O/S pups were used as
controls to test for any nonspecific effects of repeated odor
exposure training.
In the dorsolateral region, the OLR of mitral cell ensembles in
the spared OB in the O/S rats was significantly greater (49.01
0.79%) than in the occluded bulb (24.56 1.48%, n 4, t
24.84, p 1.43E 4; Fig. 2B1,B2). After associative learning, mitral cells are activated more reliably by peppermint odor, and the
same cell is likely to respond to both episodes of peppermint.
Interestingly, the total number of Arc cells activated by two odor
events did not change in the spared bulb (11.58 1.39%) compared with the occluded one (12.05 1.72%, n 4, t 0.276,
p 0.80; Fig. 2B3). However, double-stained Arc cells were
significantly increased after O/S learning (5.67 0.69% in the
spared bulb vs 3.01 0.57% in the occluded bulb; n 4, t 4.29,
p 0.02; Fig. 2B3). The percentage of single-stained Arc cells

J. Neurosci., November 12, 2014 34(46):15394 15401 15397

responding to either episode of peppermint showed a trend toward decreasing in the spared OB but did not reach statistical
significance (5.91 0.72% in the spared bulb vs 9.04 1.19% in
the occluded bulb; n 4, t 2.75, p 0.07). The increase in
double cells that are likely strongly activated by peppermint suggests odor preference learning in rat pups results in the potentiation of previously weakly activated cells.
The OLR for O/S rats was not different between the two
bulbs (spared, 33.65 0.93%; occluded, 34.22 2.42%; n 3,
t 0.17, p 0.88; Fig. 2C1,C2), suggesting no effect of odor
exposure itself on initial odor ensemble representation. Consistently, no differences were observed in the numbers of cells expressing Arc in any compartment (Fig. 2C3).
Peppermint representation in the ventromedial OB revealed
the same trends. In the O/S pups, the OLR of mitral cell ensembles was greater in the spared OB (45.07 3.59%) than in the
occluded bulb (24.40 2.22%; n 4, t 5.49, p 0.01; Fig.
2D1,D2). Consistent with the dorsolateral region, the doublestained Arc cells increased after O/S learning (6.26 1.52% in
the spared bulb vs 2.57 0.34% in the occluded bulb; n 4, t
3.07, p 0.05; Fig. 2D3), whereas the total Arc cells and singlestained Arc cells were not different in the two bulbs (Fig. 2D3).
In O/S pups, neither the OLR of cell ensembles (Fig. 2E1,E2)
nor the numbers of Arc cells (Fig. 2E3) are different in the
ventromedial OB.
An unexpected outcome was a significant reduction in the
OLR of the peppermint representation in the occluded OB in the
O/S group compared with that in the O/S group (t 3.60, p
0.02, unpaired t test). This may relate to a backward conditioning
effect when the naris plug was removed and residual peppermint
odor remained on the pup. Such an effect might be expected to
reduce the stability of peppermint encoding.
Mitral cell ensemble stabilization is specific to the
conditioned odor
In another set of experiments, we examined dorsolateral OB
Arc mitral cell ensembles to vanillin after O/S training with
peppermint (Fig. 2F ). The OLR (29.48 1.71% in the spared
bulb vs 33.14 0.85% in the occluded bulb; n 5, t 2.19, p
0.10; Fig. 2F1,F2) and the pattern of Arc expression (Fig. 2F3)
were not different between the spared and occluded bulbs. This
demonstrates that odor learning is input specific in the OB such
that only the representation of the conditioned odor is altered.
Odor preference training also results in a more stable odor
representation in the underlying granule cells of the OB
We next compared the granule cell ensembles in the OB granule
cell layer after O/S training. The areas of interest were taken
from the same rectangle regions where we measured cell ensembles in the mitral cell layers. Granule cell ensembles in the dorsolateral region showed greater OLR in the spared OB (48.07
2.99) compared with the occluded OB (24.40 2.43; n 4, t
4.56, p 0.02; Fig. 3A1,A2). The total Arc cells (7.25 0.32% in
the spared bulb vs 8.31 1.60% in the occluded bulb; n 4, t
0.60, p 0.59) and double-stained Arc cells (7.25 0.32% in
the spared bulb vs 8.31 1.60% in the occluded bulb; n 4, t
0.60, p 0.59; Fig. 3A3) were not different in the two OBs.
However, the single-stained Arc cells showed a trend of decreased numbers in the spared OB (3.76 0.21) compared with
the occluded OB (6.20 1.01; n 4, t 2.67, p 0.076; Fig.
3A3). There were no differences in either OLR or Arc cell numbers in the ventromedial region of the OB (Fig. 3B1B3). Changes
in granule cell ensembles are also training odor specific. Neither

15398 J. Neurosci., November 12, 2014 34(46):15394 15401

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

the OLR nor numbers of Arc cells were

different in the spared and occluded OB in
the dorsolateral regions to the control
odor vanillin (Fig. 3C1C3).
A more stable odor map in the aPC
We have previously shown that the OB
and the aPC are both involved in, and
support, early odor preference learning
(Lethbridge et al., 2012; Yuan and Harley,
2012; Fontaine et al., 2013; Morrison et
al., 2013). We examined pyramidal cell
ensemble changes in the aPC after early
odor preference learning from the same
animals as in the OB experiments. Singleodor exposure activates 1% pyramidal
cells in the aPC. Similar to mitral cell ensembles in the OB, the stability of the odor
representation as indexed by the OLR of
pyramidal ensembles in the spared aPC
(35.74 2.38%) was significantly greater
than that in the occluded one (18.44
2.62%; n 4, t 7.84, p 0.004; Fig.
4B1,B2). The increase in the overlap ratio
was caused by an increased number of
double-stained Arc pyramidal cells Figure 3. Early odor preference learning stabilizes the granule cell ensemble to the conditioned odor in the OB. A1A3, O/S
(0.75 0.12% in the spared hemisphere training leads to increased overlap of granule cell ensembles in the dorsolateral olfactory bulb responding to 2 peppermint
vs 0.45 0.12% in the occluded side; n exposures. A1, Example images of the granule cell layer in the occluded and spared olfactory bulbs from the same animal. A2, OLR
4, t 3.45, p 0.04), whereas the total of granule cell ensembles responding to 2 peppermint exposures. A3, Percentage of Arc cells over the total population
number of Arc cells to two odor events indexed by DAPI staining. B1B3, O/S training does not change the OLR of granule cell ensembles at the ventromedial OB

and the single-stained Arc cells were not responding to 2 peppermint exposures. C1C3, O/S training with peppermint does not change the OLR of granule cell
GC, Granule cell; DL, dorsolateral; VM, ventromedial; PP,
different in two hemispheres (Fig. 4B3). ensembles at the dorsolateral OB responding to 2 vanillin exposures.

Odor experience alone did not alter either peppermint; VA, vanillin. Arrows indicate double-stained Arc cells. Scale bars, 20 m. *p 0.05.
the overlap of the two peppermint ensemrepresentation (1500 cells) is well above the calculated threshbles (Fig. 4C1,C2) or the numbers of Arc cells activated (Fig.
old of 500 piriform pyramidal cells required to reliably drive odor
4C3).
preference behavior (Choi et al., 2011) and identical to the percentage of Kenyon cells estimated to underlie odor ensembles in
Discussion
the mushroom body of the insect (Campbell et al., 2013). It
The nature of representations
would be interesting if ensemble size was conserved in nervous
Cortical representations are known to be both sparse, reflecting a
system evolution.
dynamic balance of excitatory and inhibitory inputs, and variable
A curious aspect of the present data is the finding of an 15
(Shadlen and Newsome, 1998; Olshausen and Field, 2004). These
20% overlap among unrelated odors, which is substantially larger
characteristics are thought to account for the large storage capacthan a 0.5% overlap that would be predicted from the size of each
ity of mammalian brain and reflect the dynamic aspects of its
odors representation (7%) by random draw with replacement.
operation. Although representation in the OB itself is more like
We suggest that this overlap reflects the existence of an active
that of sensory cortices in having a spatial organization such that
subset of cortical neurons that are primed to participate in any
we are able to target representational regions, the aPC behaves like
representation in a given time window (Yassin et al., 2010; Luczak
the general associative cortical model (Johnson et al., 2000). Comand Maclean, 2012; Mizuseki and Buzsaki, 2013; Klinshov et al.,
pared with adult aPC (Shakhawat et al., 2014), the odor ensembles
2014). Such primed subsets require a reconfiguration of our norin rat pup aPC were significantly smaller (3% vs 1%). We
mal
thinking about distributed random neural networks.
suggest this difference relates directly to the maturation of lateral
The first conclusion that can be made about granule cell parolfactory tract input to the piriform cortex, which is about oneticipation in odor ensembles from these data is that it appears to
third of the adult value at this age (Sarma et al., 2011). Earlier
be odor specific, arguing for different mitral cell/granule cell enestimates of piriform ensemble size have been substantially larger
sembles for different odors. Johnson and Leon (1996), using
(Poo and Isaacson, 2009; Stettler and Axel, 2009), but this is likely
2-deoxyglucose (2-DG), showed that peppermint activates two
a function of probing ensembles in the anesthetized versus awake state
hot spots in the glomerular layer of the OB, one in the dorsolat(Kato et al., 2012). The present values derive from ensemble measureeral region and one in the ventromedial region. Early preference
ments in awake animals.
learning predominately enhanced 2-DG activation in the dorsoRat pups have similar numbers of piriform pyramidal cells as
lateral glomerular region (Johnson and Leon, 1996) and phosthe adult (Sarma et al., 2011), and so dividing total piriform
phorylated CREB in the dorsolateral mitral cell layer (McLean et
stereological counts (Capurso et al., 1997; Duffell et al., 2000) in
al., 1999), consistent with the more prominent change in the Arc
half provides an estimate of 150,000 cells available in each
hemisphere to participate in aPC representations. Thus, a 1%
expression of mitral and granule cells in dorsolateral region. In

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

J. Neurosci., November 12, 2014 34(46):15394 15401 15399

Figure 4. Early odor preference learning stabilizes the odor map for the conditioned odor in the aPC. A, Schematic of aPC and Arc sampling regions (red rectangles). B1B3, O/S training leads
to increased overlap of pyramidal cell ensembles in the aPC responding to 2 peppermint exposures. B1, Example images of the pyramidal cell layer in the occluded and spared aPC from the same
animal. B2, OLR of pyramidal cell ensembles responding to 2 peppermint exposures. B3, Percentage of Arc cells over the total population indexed by DAPI staining. C1C3, O/S training does
not change the OLR of pyramidal cell ensembles in the aPC responding to 2 peppermint exposures. PP, Peppermint. Arrows indicate double-stained Arc cells. Scale bars, 20 m. *p 0.05;
**p 0.01.

19-d-old rats, c-Fos granule cells significantly decrease with odor

learning (Woo et al., 1996), as do Arc pyramidal cells in piriform cortex of adult rats (Shakhawat et al., 2014). The lack of a
decrease here in either area is likely related to age.
The second conclusion is that, like mitral cells, the granule cell
representation of an odor increases its stability after the pairing
with stroking reward. This parallel change in the granule cell and
mitral cell ensembles is consistent with the idea that changes in
excitation in any cortical system will be accompanied by balanced
inhibition (Isaacson and Scanziani, 2011; Saar et al., 2012; Xue et
al., 2014). Mitral cells driving granule cells provides the most
parsimonious account of these effects, and if that is the case, it
again underlines a highly selective relationship among mitral and
granule cells. Consistent with such selectivity, electrical coupling
between mitral cells and nearby underlying granule cells has been
reported in rat pups (Paternostro et al., 1995). Feedback effects
from aPC that drive granule cell inhibition for individual odors
has also been demonstrated (Restrepo et al., 2009; Boyd et al.,
2012) and is another possible source of support for the parallel
stability increases observed in the granule cell ensembles.
Overall, the striking feature of the learning-related changes in
odor representation observed in these experiments is the increase
in the stability of ensemble representation from 25% to 49% in
the OB and from 18% to 35% in the aPC. The level of overlap
after our odor reward pairings in the OB is similar to what has
been observed using Arc to identify representations of repeated
strong visual input in secondary visual cortex (50% overlap; Rudinskiy et al., 2012). This similarity of overlap levels in sensory
stimuli for rewarded odor and for strong visual stimulation is
consistent with data showing odor learning modifies OB responses to be similar to responses to a higher concentration of

odorant (Abraham et al., 2014). Recent modeling work on cortical system representations argues that the stability parameter in
population vectors is critical for adaptive behavior (Montijn et
al., 2014). These changes in the responses to simple odorants were
not able to be previously characterized using electrophysiological
methods to probe representations (Chapuis and Wilson, 2012).
Generality of the rat pup model
There are many parallels between the rat pup odor preference
model and adult odor associative learning models. Adult aPC
ensembles also show the stabilization effect of learning and memory, but the number of neurons participating in an ensemble
becomes somewhat sparser than before learning (Shakhawat et
al., 2014), whereas that number did not change in the rat pup. In
neither model does enlargement of the rewarded representation
occur; this is consistent with data suggesting enlargement of sensory representations does not account for long-term memory
even when it is seen (Reed et al., 2011). However, multiple groups
have reported enlarged OB glomerular representations with
learning in both rodent pups and adults (Woo et al., 1987; Johnson and Leon, 1996; Abraham et al., 2014). We have also described such a glomerular effect using intrinsic optical imaging in
the odor preference learning model (Yuan et al., 2002), and as
mentioned earlier, these effects are similar to those of increasing
concentrations of the odorant (Abraham et al., 2014). Both enhanced glomerular input and increased stability of principal neuronal network representations should serve to create a stronger
and more discriminable experiential input.
The machinery for NE to act as an unconditioned stimulus in
the rat pup (Yuan et al., 2014) remains in the OB of older rats, and
recent data suggest that blocking both - and -adrenoceptors

15400 J. Neurosci., November 12, 2014 34(46):15394 15401

in the adult OB prevents discrimination of similar odors

(Doucette et al., 2007; Mandairon et al., 2008). Whereas NE
via -adrenoceptor activation also mediates early odor learning
in rat pup aPC (Morrison et al., 2013), as in rat pup OB, the role
of NE projections and the function of NE in the aPC in adult rat
odor learning requires future investigation. NMDARs and L-type
calcium channels are critical in the OB as calcium sources mediating plasticity (Jerome et al., 2012; Lethbridge et al., 2012), and
they are likely involved in aPC plasticity and aPC-mediated learning (Morrison et al., 2013), as both NMDARs and L-type calcium
channels are critical mediators for Arc activation in the hippocampus (Bateup et al., 2013).
The cellular and intracellular supports of rat pup learning and
memory (see Introduction) are also those implicated in invertebrate and vertebrate associative learning and appear central for
learning and memory in mammalian brain across the life span.
Arc and plasticity
Arc here identifies the neurons participating in responding to
odors, with the advantage of capturing the ensembles to the same
odor twice. Neurons either alter their firing rate or increase their
firing reliability to an odor after learning (Doucette et al., 2011).
Increases in neuronal firing reliability translate into a tighter
overlap in the condition in which an animal receives the same
odor twice. Our data from both rat pups and adults (Shakhawat et
al., 2014) suggest that the probability that weakly activated cells
transcribe Arc twice is lower before conditioning than after.
Arc is also part of the plasticity story. Others have suggested
that CREB and/or immediate-early genes like Arc identify neurons that are primed to participate in memory ensembles (Han et
al., 2007; Yiu et al., 2014). Arc has recently been shown to promote thin spine production as sites for connectivity strengthening while homeostatically downregulating weaker connections
(Peebles et al., 2010). Arc-negative mice show neither depression
nor potentiation as a function of visual experience, whereas with
olfactory experience, both potentiation and depression operate
and are hypothesized to modulate the aPC ensemble changes
across pups and adults (Saar et al., 2012; Yuan et al., 2014). Thin
spine growth, as reported in hippocampus, may contribute to
lasting ensemble strengthening in aPC, as well as among OB
granule cells.

References
Abraham NM, Vincis R, Lagier S, Rodriguez I, Carleton A (2014) Long term
functional plasticity of sensory inputs mediated by olfactory learning.
Elife (Cambridge) 3:e02109. CrossRef Medline
Bateup HS, Denefrio CL, Johnson CA, Saulnier JL, Sabatini BL (2013) Temporal dynamics of a homeostatic pathway controlling neural network
activity. Front Mol Neurosci 6:28. CrossRef Medline
Boyd AM, Sturgill JF, Poo C, Isaacson JS (2012) Cortical feedback control of
olfactory bulb circuits. Neuron 76:11611174. CrossRef Medline
Campbell RA, Honegger KS, Qin H, Li W, Demir E, Turner GC (2013)
Imaging a population code for odor identity in the Drosophila mushroom
body. J Neurosci 33:10568 10581. CrossRef Medline
Capurso SA, Calhoun ME, Sukhov RR, Mouton PR, Price DL, Koliatsos VE
(1997) Deafferentation causes apoptosis in cortical sensory neurons in
the adult rat. J Neurosci 17:73727384. Medline
Chapuis J, Wilson DA (2012) Bidirectional plasticity of cortical pattern recognition and behavioral sensory acuity. Nat Neurosci 15:155161.
CrossRef Medline
Choi GB, Stettler DD, Kallman BR, Bhaskar ST, Fleischmann A, Axel R
(2011) Driving opposing behaviors with ensembles of piriform neurons.
Cell 146:1004 1015. CrossRef Medline
Cui W, Smith A, Darby-King A, Harley CW, McLean JH (2007) A temporalspecific and transient cAMP increase characterizes odorant classical conditioning. Learn Mem 14:126 133. CrossRef Medline

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

Cui W, Darby-King A, Grimes MT, Howland JG, Wang YT, McLean JH,
Harley CW (2011) Odor preference learning and memory modify
GluA1 phosphorylation and GluA1 distribution in the neonate rat olfactory bulb: testing the AMPA receptor hypothesis in an appetitive learning
model. Learn Mem 18:283291. CrossRef Medline
Doucette W, Milder J, Restrepo D (2007) Adrenergic modulation of olfactory bulb circuitry affects odor discrimination. Learn Mem 14:539 547.
CrossRef Medline
Doucette W, Gire DH, Whitesell J, Carmean V, Lucero MT, Restrepo D
(2011) Associative cortex features in the first olfactory brain relay station.
Neuron 69:1176 1187. CrossRef Medline
Duffell SJ, Soames AR, Gunby S (2000) Morphometric analysis of the developing rat brain. Toxicol Pathol 28:157163. CrossRef Medline
Fontaine CJ, Harley CW, Yuan Q (2013) Lateralized odor preference training in rat pups reveals an enhanced network response in anterior piriform
cortex to olfactory input that parallels extended memory. J Neurosci 33:
15126 15131. CrossRef Medline
Grimes MT, Harley CW, Darby-King A, McLean JH (2012) PKA increases
in the olfactory bulb act as unconditioned stimuli and provide evidence
for parallel memory systems: pairing odor with increased PKA creates
intermediate- and long-term, but not short-term, memories. Learn Mem
19:107115. CrossRef Medline
Guzowski JF, Worley PF (2001) Cellular compartment analysis of temporal
activity by fluorescence in situ hybridization (catFISH). Curr Protoc Neurosci Chapter 1:Unit 1 8. CrossRef Medline
Guzowski JF, Timlin JA, Roysam B, McNaughton BL, Worley PF, Barnes CA
(2005) Mapping behaviorally relevant neural circuits with immediateearly gene expression. Curr Opin Neurobiol 15:599 606. CrossRef
Medline
Han JH, Kushner SA, Yiu AP, Cole CJ, Matynia A, Brown RA, Neve RL,
Guzowski JF, Silva AJ, Josselyn SA (2007) Neuronal competition and
selection during memory formation. Science 316:457 460. CrossRef
Medline
Harley CW, Darby-King A, McCann J, McLean JH (2006) Beta1adrenoceptor or alpha1-adrenoceptor activation initiates early odor preference learning in rat pups: support for the mitral cell/cAMP model of
odor preference learning. Learn Mem 13:8 13. CrossRef Medline
Isaacson JS, Scanziani M (2011) How inhibition shapes cortical activity.
Neuron 72:231243. CrossRef Medline
Jerome D, Hou Q, Yuan Q (2012) Interaction of NMDA receptors and
L-type calcium channels during early odor preference learning in rats. Eur
J Neurosci 36:3134 3141. CrossRef Medline
Johnson BA, Leon M (1996) Spatial distribution of [14C]2-deoxyglucose
uptake in the glomerular layer of the rat olfactory bulb following early
odor preference learning. J Comp Neurol 376:557566. CrossRef Medline
Johnson DM, Illig KR, Behan M, Haberly LB (2000) New features of connectivity in piriform cortex visualized by intracellular injection of pyramidal cells suggest that primary olfactory cortex functions like
association cortex in other sensory systems. J Neurosci 20:6974 6982.
Medline
Kato HK, Chu MW, Isaacson JS, Komiyama T (2012) Dynamic sensory
representations in the olfactory bulb: modulation by wakefulness and
experience. Neuron 76:962975. CrossRef Medline
Klinshov VV, Teramae JN, Nekorkin VI, Fukai T (2014) Dense neuron clustering explains connectivity statistics in cortical microcircuits. PLoS One
9:e94292. CrossRef Medline
Kucharski D, Hall WG (1987) New routes to early memories. Science 238:
786 788. CrossRef Medline
Kucharski D, Johanson IB, Hall WG (1986) Unilateral olfactory conditioning in 6-day-old rat pups. Behav Neural Biol 46:472 490. CrossRef
Medline
Lethbridge R, Hou Q, Harley CW, Yuan Q (2012) Olfactory bulb glomerular NMDA receptors mediate olfactory nerve potentiation and odor preference learning in the neonate rat. PLoS One 7:e35024. CrossRef Medline
Logan DW, Brunet LJ, Webb WR, Cutforth T, Ngai J, Stowers L (2012)
Learned recognition of maternal signature odors mediates the first suckling episode in mice. Curr Biol 22:1998 2007. CrossRef Medline
Luczak A, Maclean JN (2012) Default activity patterns at the neocortical
microcircuit level. Front Integr Neurosci 6:30. CrossRef Medline
Lyford GL, Yamagata K, Kaufmann WE, Barnes CA, Sanders LK, Copeland
NG, Gilbert DJ, Jenkins NA, Lanahan AA, Worley PF (1995) Arc, a
growth factor and activity-regulated gene, encodes a novel cytoskeleton-

Shakhawat, Gheidi et al. Arc Visualization of Odor Memory

associated protein that is enriched in neuronal dendrites. Neuron 14:433
445. CrossRef Medline
Mandairon N, Peace S, Karnow A, Kim J, Ennis M, Linster C (2008) Noradrenergic modulation in the olfactory bulb influences spontaneous and
reward-motivated discrimination, but not the formation of habituation
memory. Eur J Neurosci 27:1210 1219. CrossRef Medline
McCurry CL, Shepherd JD, Tropea D, Wang KH, Bear MF, Sur M (2010)
Loss of Arc renders the visual cortex impervious to the effects of sensory
experience or deprivation. Nat Neurosci 13:450 457. CrossRef Medline
McLean JH, Harley CW, Darby-King A, Yuan Q (1999) pCREB in the neonate rat olfactory bulb is selectively and transiently increased by odor
preference-conditioned training. Learn Mem 6:608 618. CrossRef
Medline
Miyashita T, Kubik S, Haghighi N, Steward O, Guzowski JF (2009) Rapid
activation of plasticity-associated gene transcription in hippocampal neurons provides a mechanism for encoding of one-trial experience. J Neurosci 29:898 906. CrossRef Medline
Mizuseki K, BuzsakiG (2013) Preconfigured, skewed distribution of firing
rates in the hippocampus and entorhinal cortex. Cell Rep 4:1010 1021.
CrossRef Medline
Montijn JS, Vinck M, Pennartz CM (2014) Population coding in mouse
visual cortex: response reliability and dissociability of stimulus tuning and
noise correlation. Front Comput Neurosci 8:58. CrossRef Medline
Morrison GL, Fontaine CJ, Harley CW, Yuan Q (2013) A role for the anterior
piriform cortex in early odor preference learning: evidence for multiple olfactory learning structures in the rat pup. J Neurophysiol 110:141152. CrossRef
Medline
Olshausen BA, Field DJ (2004) Sparse coding of sensory inputs. Curr Opin
Neurobiol 14:481 487. CrossRef Medline
Paternostro MA, Reyher CK, Brunjes PC (1995) Intracellular injections of
Lucifer yellow into lightly fixed mitral cells reveal neuronal dye-coupling
in the developing rat olfactory bulb. Brain Res Dev Brain Res 84:110.
CrossRef Medline
Peebles CL, Yoo J, Thwin MT, Palop JJ, Noebels JL, Finkbeiner S (2010) Arc
regulates spine morphology and maintains network stability in vivo. Proc
Natl Acad Sci U S A 107, 1817318178. CrossRef Medline
Poo C, Isaacson JS (2009) Odor representations in olfactory cortex: sparse
coding, global inhibition, and oscillations. Neuron 62:850 861. CrossRef
Medline
Rangel S, Leon M (1995) Early odor preference training increases olfactory
bulb norepinephrine. Brain Res Dev Brain Res 85:187191. CrossRef
Medline
Reed A, Riley J, Carraway R, Carrasco A, Perez C, Jakkamsetti V, Kilgard MP
(2011) Cortical map plasticity improves learning but is not necessary for
improved performance. Neuron 70:121131. CrossRef Medline
Restrepo D, Doucette W, Whitesell JD, McTavish TS, Salcedo E (2009)
From the top down: flexible reading of a fragmented odor map. Trends
Neurosci 32:525531. CrossRef Medline
Rudinskiy N, Hawkes JM, Betensky RA, Eguchi M, Yamaguchi S, Spires-Jones
TL, Hyman BT (2012) Orchestrated experience-driven Arc responses
are disrupted in a mouse model of Alzheimers disease. Nat Neurosci
15:14221429. CrossRef
Rumsey JD, Darby-King A, Harley CW, McLean JH (2001) Infusion of the
metabotropic receptor agonist, DCG-IV, into the main olfactory bulb
induces olfactory preference learning in rat pups. Brain Res Dev Brain Res
128:177179. CrossRef Medline

J. Neurosci., November 12, 2014 34(46):15394 15401 15401

Saar D, Reuveni I, Barkai E (2012) Mechanisms underlying rule learninginduced enhancement of excitatory and inhibitory synaptic transmission.
J Neurophysiol 107:12221229. CrossRef
Sarma AA, Richard MB, Greer CA (2011) Developmental dynamics of piriform cortex. Cereb Cortex 21:12311245. CrossRef Medline
Shadlen MN, Newsome WT (1998) The variable discharge of cortical neurons: implications for connectivity, computation, and information coding. J Neurosci 18:3870 3896. Medline
Shakhawat AM, Harley CW, Yuan Q (2012) Olfactory bulb alpha2adrenoceptor activation promotes rat pup odor-preference learning via a
cAMP-independent mechanism. Learn Mem 19:499 502. CrossRef
Medline
Shakhawat AM, Harley CW, Yuan Q (2014) Arc visualization of odor objects reveals experience-dependent ensemble sharpening, separation, and
merging in anterior piriform cortex in adult rat. J Neurosci 34:10206
10210. CrossRef Medline
Stettler DD, Axel R (2009) Representations of odor in the piriform cortex.
Neuron 63:854 864. CrossRef Medline
Sullivan RM, Stackenwalt G, Nasr F, Lemon C, Wilson DA (2000) Association of an odor with activation of olfactory bulb noradrenergic betareceptors or locus coeruleus stimulation is sufficient to produce learned
approach responses to that odor in neonatal rats. Behav Neurosci 114:
957962. CrossRef Medline
Vazdarjanova A, Ramirez-Amaya V, Insel N, Plummer TK, Rosi S, Chowdhury S, Mikhael D, Worley PF, Guzowski JF, Barnes CA (2006) Spatial
exploration induces ARC, a plasticity-related immediate-early gene, only
in calcium/calmodulin-dependent protein kinase II-positive principal excitatory and inhibitory neurons of the rat forebrain. J Comp Neurol 498:
317329. CrossRef Medline
Woo CC, Coopersmith R, Leon M (1987) Localized changes in olfactory
bulb morphology associated with early olfactory learning. J Comp Neurol
263:113125. CrossRef Medline
Woo CC, Oshita MH, Leon M (1996) A learned odor decreases the number
of Fos-immunopositive granule cells in the olfactory bulb of young rats.
Brain Res 716:149 156. CrossRef Medline
Xue M, Atallah BV, Scanziani M (2014) Equalizing excitation-inhibition
ratios across visual cortical neurons. Nature 511:596 600. CrossRef
Medline
Yassin L, Benedetti BL, Jouhanneau JS, Wen JA, Poulet JF, Barth AL (2010)
An embedded subnetwork of highly active neurons in the neocortex.
Neuron 68:10431050. CrossRef Medline
Yiu AP, Mercaldo V, Yan C, Richards B, Rashid AJ, Hsiang HL, Pressey J,
Mahadevan V, Tran MM, Kushner SA, Woodin MA, Frankland PW,
Josselyn SA (2014) Neurons are recruited to a memory trace based on
relative neuronal excitability immediately before training. Neuron 83:
722735. CrossRef Medline
Yuan Q, Harley CW (2012) What a nostril knows: olfactory nerve-evoked
AMPA responses increase while NMDA responses decrease at 24-h posttraining for lateralized odor preference memory in neonate rat. Learn
Mem 19:50 53. CrossRef Medline
Yuan Q, Harley CW, McLean JH, Knopfel T (2002) Optical imaging of odor
preference memory in the rat olfactory bulb. J Neurophysiol 87:3156
3159. CrossRef Medline
Yuan Q, Shakhawat AM, Harley CW (2014) Mechanisms underlying early
odor preference learning in rats. Prog Brain Res 208:115156. CrossRef
Medline