precise circuitry links bilaterally symmetric olfactory maps

Neuron

Article

Precise Circuitry Links BilaterallySymmetric Olfactory MapsZhiqiang Yan,1,2 Jie Tan,1,2 Chang Qin,2 Yao Lu,2 Cheng Ding,2 and Minmin Luo2,*1Institute of Biophysics, Chinese Academy of Sciences, Beijing, 100101, China2National Institute of Biological Sciences, Beijing, 102206, China

*Correspondence: [email protected] 10.1016/j.neuron.2008.03.012

SUMMARY

Olfactory sensory neurons expressing a commonreceptor gene converge onto one or a few glomeruliwith stereotyped positions within the mouse mainolfactory bulb (MOB), producing a map of �1800olfactory columns representing �1000 odorantreceptors. Here, we report that this precise olfactorymap is maintained over several synapses thatultimately cross MOB hemispheres to link bilateralisofunctional olfactory columns. Focal injection oftracer into genetically identified glomeruli revealedan exquisite topography that involves a bilateral con-nection via the anterior olfactory nucleus parsexterna (AONpE) that links isofunctional olfactorycolumns in the contralateral MOB. Physiologicaland behavioral assays revealed an important rolefor the AONpE in bilateral exchange of odorant-spe-cific information. These results indicate that the inter-bulbar link through the AONpE integrates bilateralolfactory sensory maps and exchanges olfactoryinformation, positioning it as a unique model systemfor studying interhemispheric connections.

INTRODUCTION

All sensory modalities utilize topographically organized sensory

maps to represent features of sensory stimuli (Luo and Flanagan,

2007; O’Leary et al., 1999). Central topography typically reflects

the initial organization of the sensory epithelium, in which recep-

tor neurons are spatially organized based on their types and tun-

ing (Luo and Flanagan, 2007; O’Leary et al., 1999). The olfactory

system is unusual in this regard because olfactory sensory neu-

rons (OSNs) expressing a given receptor gene out of a repertoire

of �1000 are scattered in the olfactory epithelium (Buck and

Axel, 1991; Mombaerts et al., 1996; Ressler et al., 1994; Vassar

et al., 1994; Zhang and Firestein, 2002). This disarray is dramat-

ically transformed to ordered organization at the level of glomer-

ulus in the MOB. Here, OSNs expressing a common receptor

gene converge precisely into one or two glomeruli with a stereo-

typed location in the ipsilateral MOB (Mombaerts et al., 1996;

Ressler et al., 1994; Vassar et al., 1994). In mice, each MOB

has�1800 glomeruli, and in each glomerulus OSNs form synap-

ses with a specific set of mitral/tufted cells (Shepherd et al.,

2004). Like columns in the cerebral cortex, a glomerulus forms

part of a functional unit, or ‘‘olfactory column,’’ consisting of

mitral/tufted cells, periglomerular cells, and underlying granule

cells (Shepherd et al., 2004; Willhite et al., 2006). Isofunctional

olfactory columns are those that are associated with the same

odorant receptors. Each mouse MOB can thus be viewed as

a spatial map containing �1800 olfactory columns with parallel

inputs and outputs.

The stereotyped position of identified glomeruli is preserved

not only between individuals but also between hemispheres

within an individual, creating bilaterally symmetric olfactory

sensory maps (Mombaerts et al., 1996; Ressler et al., 1994;

Rubin and Katz, 1999; Uchida et al., 2000; Vassar et al.,

1994; Wachowiak and Cohen, 2001). Although bilateral olfac-

tory bulbs lack direct connections between them, they can in-

teract indirectly via feedback pathways. A cortical area called

the anterior olfactory nucleus (AON) is known to contain neu-

rons that project to the contralateral MOB (Davis and Macrides,

1981; Haberly and Price, 1978; Schoenfeld and Macrides,

1984; Scott et al., 1985). Within the AON, the anterior olfactory

nucleus pars externa (AONpE) receives input from the ipsilat-

eral MOB and projects exclusively to the granule cell layer of

the contralateral MOB, forming the interbulbar association sys-

tem (Davis and Macrides, 1981; Schoenfeld and Macrides,

1984; Scott et al., 1985). Many important questions remain un-

addressed about the organization of this system. In particular,

do bilateral olfactory maps interact with any specificity at the

level of odorant receptors, and is there a functional role if

such interactions exist?

We studied these questions by utilizing genetically engineered

mice that express green fluorescence protein (GFP) in a specific

set of OSNs that also express the endogenous M71 receptor.

This receptor has a characterized ligand; thus, we were able to

probe both anatomical and functional organization. By targeting

tracers to genetically identified glomeruli, we uncovered an as yet

unknown level of organization in the olfactory system, where

bilateral isofunctional olfactory columns are precisely linked by

neurons in the AONpE. We found that odorant-specific activation

of the contralateral MOB required the AONpE. In addition, we

developed a behavioral assay to establish the important role of

the AONpE in crosslateral transfer of odorant-specific olfactory

memory. These results thus demonstrate that the precision of

the olfactory sensory map is preserved along interhemispheric

circuitry and that these connections contribute to bilateral

exchange of specific olfactory information, including memorized

olfactory associations.

Neuron 58, 613–624, May 22, 2008 ª2008 Elsevier Inc. 613

mailto:[email protected]

Neuron

Precise Linking of Bilateral Olfactory Maps

RESULTS

The Interbulbar Association System Links BilateralIsofunctional Olfactory ColumnsTo target tract tracers into a single identified olfactory glomeru-

lus, we first utilized a genetically engineered mouse line in which

GFP is coexpressed with the odorant receptor M71 (Bozza et al.,

2002). In these mice, the lateral GFP-labeled M71 glomerulus is

typically located in the dorsoposterior surface of the MOB.

Guided by GFP fluorescence, we made focal iontophoretic injec-

tions of fluorescence-conjugated dextran amines into the lateral

Figure 1. The Projection from the MOB to the Ipsilat-

eral AONpE Is Topographic

(A) Schematic representation of experimental strategy and re-

sulting anterograde labeling. Focal deposit of tract tracer (red

pipette) into a dorsal M71 glomerulus (green) anterogradely

labels a small patch of terminals in the dorsal ipsilateral AONpE.

(B) Focal injection of red tracer into a single dorsal M71 glo-

merulus (green). Inset shows a high-power view of the injec-

tion site (arrow). Glomeruli were delineated by DAPI labeling

(blue).

(C) This injection anterogradely labeled a small patch of axo-

nal terminals (arrow) in layer Ia of the dorsal ipsilateral AONpE

(dashed lines). Inset shows a high-power view of axonal label-

ing (arrow).

(D) Schematic representation of results of injecting dyes into

the mitral/tufted cells in the ventral MOB. Same conventions

as in (A).

(E and F) Injection into the ventral layer of mitral/tufted cells (E)

labeled a patch of axon terminals confined to the ventral

AONpE.

(G–I) Topography along the anterior-posterior (A-P) axis for

the MOB-to-AONpE projection. (G) Coronal sections showing

injection sites in the glomerular layer of the anterior (red, top

panel) and posterior (green, bottom panel) sections of an

MOB. Distance between the two sections was�720 mm along

the A-P axis. (H and I) Anterograde labeling in the anterior

AONpE was mostly red (H) whereas that in posterior AONpE

was mostly green (I). Distance between the two sections

was �120 mm along the A-P axis.

In this and following figures, all sections are coronal. GL, glo-

merular layer; MCL, mitral cell layer; GCL, granule cell layer.

Scale bars, 200 mm.

M71 glomerulus (Figures 1A and 1B; Figure S1A,

available online). In all cases, we observed a single

patch of intensely labeled axon terminals within

layer Ia of the dorsal AONpE (Figure 1C; n = 7

mice; see Figure S2 for topology of the AONpE).

The location of the axonal patches in the AONpE

was highly stereotyped across animals and was

spatially restricted, spanning only approximately

100 mm at levels approximately 3.2 mm anterior

to Bregma. These results indicate specificity in

the projections from the MOB to the ipsilateral

AONpE. To examine the topographic relationship

between the MOB and AONpE further, we made

tracer injections into different glomeruli in the

MOB (Figure 1; Figure S3). When tracers were in-

jected into the mitral/tufted cells in the ventral

MOB, we observed a distinct cluster of anterogradely labeled

axonal terminals in the ventral AONpE (Figures 1D–1F; n = 5

mice). Tracer injections into the dorsomedial glomeruli near the

medial M71 glomerulus anterogradely labeled a cluster of axonal

terminals within the dorsomedial AONpE (Figures S3A and S3B;

n = 4 mice). Injections into glomeruli located in the middle level of

the medial MOB labeled a small cluster of terminals in ventrome-

dial AONpE (Figures S3C and S3D; n = 3 mice). We next injected

two tracers with different fluorophore into two distinct positions

along the anterior-posterior axis of the dorsal MOB. Anterog-

radely labeled terminal fields were largely separated in the dorsal

614 Neuron 58, 613–624, May 22, 2008 ª2008 Elsevier Inc.

Neuron


AONpE (Figures 1G–1I; Figure 2G; n = 8 mice, r = 0.97, p < 0.001;

Pearson’s correlation).

Having established the topography for the projection from the

MOB to ipsilateral AONpE, we next examined whether the projec-

tion from the AONpE to contralateral MOB was also topographic.

The projection from the AONpE terminates in the granule cell

layer of the contralateral MOB. We thus made small injections

into the granule cell layer and examined the distribution of retro-

gradely labeled somata in the contralateral AONpE (Figure 2).

After making focal tracer injections into a small region within the

granule cell layer below the pial surface of the lateral M71 glomer-

Figure 2. The Projection from the AONpE to the

Contralateral MOB Is Topographic

(A) Schematic representation of results of retrograde

labeling in the contralateral AONpE following tracer in-

jection into the granule cell layer of the MOB. Dashed

line indicates midline dividing hemispheres.

(B and C) Focal tracer deposit into the granule cell

layer under the dorsal M71 glomerulus (B) retrogradely

labeled somata clustered within the dorsal AONpE of

the contralateral side ([C], dashed line). Inset in (C)

shows high-power view of the labeled somata (arrow

in [C]).

(D) Schematic representation of results following

tracer injection into the ventral granule cell layer.

(E and F) Tracer deposit into the ventral granule cell

layer (E) retrogradely labeled somata clustered in the

ventral AONpE of the contralateral side (F).

(G) Along the A-P axis, the location of anterograde

labeling in the AONpE was linearly correlated with

the injection sites in the glomerular layer (red line

with squares), as was retrograde labeling in the con-

tralateral AONpE with injections into the granule cell

layer (blue line with circles). Numbers indicate the dis-

tance relative to the rostral end of the MOB or the

AONpE along the A-P axis.

Scale bars, 200 mm.

ulus, a tight cluster of somata was retro-

gradely labeled in the contralateral AONpE

(Figures 2A–2C; n = 6 mice). Across animals,

retrograde labeling consistently appeared in

the dorsal AONpE and was limited in size

to �100 mm. This area corresponded to the

location of anterograde labeling observed af-

ter tracer injection into the ipsilateral M71 glo-

merulus (Figure 1C; Figure 2C). To test the

topography of contralateral AONpE projec-

tions to areas distal from the M71 glomerulus,

we made tracer injections into the granule cell

layer in the ventral MOB and observed a small

cluster of retrogradely labeled somata in the

ventral AONpE contralateral to the injection

site (Figures 2D–2F; n = 3 mice). Focal tracer

injection into the granule cell layer in the dor-

somedial MOB near the medial M71 glomer-

ulus retrogradely labeled a cluster of somata

in the dorsomedial area of the contralateral

AONpE (Figure S4; n = 3 mice). This area cor-

responded to the location of anterograde labeling after tracer

injection into the ipsilateral dorsomedial glomeruli (Figures S3A

and S3B). Along the anterior-posterior axis, the topography in

the AONpE-to-contralateral MOB projection resembled that of

the MOB-to-ipsilateral AONpE projection (Figure 2G; n = 5

mice, r = 0.98, p < 0.001; Pearson’s correlation).

Our data thus suggest that isofunctional olfactory columns in

the bilateral olfactory bulbs are linked via their topographic con-

nections with the AONpE. To directly test this possibility, we

injected red tracer into one M71 glomerulus and green tracer

into the granule cell layer under the contralateral M71 glomerulus


Neuron


Figure 3. Precise Linking of Isofunctional Olfactory Columns of the Bilateral Olfactory Bulbs by the AONpE

(A) Schematic representation of experimental strategy and results. Red tracer was injected into the M71 glomerulus and green tracer was injected into the granule

cell layer beneath the contralateral M71 glomerulus.

(B) Deposit of red tracer into the dorsal M71 glomerulus of the left hemisphere and green tracer into the granule cell layer beneath the M71 glomerulus of the right

hemisphere. Inset in the right-hand panel shows that the injection site (red, arrow) was restricted to the M71 glomerulus (green).

(C) Following this injection, retrogradely labeled somata and anterogradely labeled terminals were colocalized in a small region of the dorsal AONpE ipsilateral to

the injection site of red tracer. Panel at right shows a high-power view of labeling (arrow at left).

(D) Schematic representation of results following focal tracer injection into the AONpE.

(E and F) Deposit of tracer into a small region within the AONpE (E) retrogradely labeled mitral/tufted cells within the dorsal MOB of the ipsilateral side and ante-

rogradely labeled terminals in the dorsal bulb of the contralateral side (F). Note the symmetry of labeling (arrows in [F]) within the bilateral bulbs. Insets at right with

numbers 1 and 2 show high-power views corresponding to numeric labels shown in left-hand panels.

(G) A highly simplified schematic diagram showing the topographic relationship for the projection from the MOB to the ipsilateral AONpE. Small circles of different

color in the left-hand panel indicate injection sites made in the MOB, whereas thatched circles with matching color in the right-hand panel indicate the corre-

sponding target areas in the AONpE. A, anterior; DL, dorsolateral; DM, dorsomedial; P, posterior; VL, ventrolateral; VM, ventromedial.


Neuron


(Figures 3A and 3B; n = 4 mice). In all four cases, we observed

colocalization of anterogradely labeled terminals and retro-

gradely labeled somata within a highly restricted region of the

AONpE ipsilateral to the injected M71 glomerulus (Figure 3C).

The dendrites of retrogradely labeled somata intermingled with

ramifying axon terminals from the ipsilateral glomerulus injection

(Figure 3C, right-hand panel). To further test whether this homo-

topic linking is specific to the M71 glomerulus alone or applies

more generally to other glomeruli, we carried out tracing exper-

iments targeting lateral mI7/M71 glomeruli in mI7/M71-

IRES-tauGFP mice (Bozza et al., 2002). In these mice, the gene

encoding the M71 receptor is replaced with the gene for the

mouse I7 receptor, and the lateral mI7/M71 glomeruli are

ectopically located in a position anterior to the endogenous lat-

eral M71 glomeruli in the dorsal MOB. Injections into the lateral

mI7/M71 glomerulus anterogradely labeled a cluster of axon

terminals in the dorsal AONpE that was more anterior to the

AONpE area targeted by the lateral M71 glomerulus (Figures

S5A–S5C; n = 4 mice). Small injections into the granule cell layer

below the lateral mI7/M71 glomerulus retrogradely labeled an

area in the contralateral AONpE apparently corresponding to the

area targeted by ipsilateral mI7/M71 glomerulus (Figures S5E

and S5F; n = 4 mice).

Further demonstrating the homotopic linking by the AONpE,

tracer deposit into small regions of the dorsal AONpE labeled

bilaterally symmetric regions in the MOB (Figures 3D–3F; n = 6

mice). Retrogradely labeled somata of mitral/tufted cells within

the ipsilateral bulb and anterogradely labeled axon terminals in

the contralateral bulb occupied highly restricted regions of the

dorsal bulb that were bilaterally symmetric (arrows in Figure 3F).

Confirming previous studies indicating that the AONpE projects

exclusively to the contralateral MOB (Davis and Macrides,

1981; Schoenfeld and Macrides, 1984), we did not observe any

anterograde labeling in cortical regions after tracer injection into

the AONpE.

OSNs expressing common receptors tend to project to just

two glomeruli, one medial and one lateral in the olfactory bulb,

forming two mirror-symmetric isofunctional glomeruli within

each MOB (Mombaerts et al., 1996; Ressler et al., 1994; Vassar

et al., 1994). The two isofunctional glomeruli form precise recip-

rocal connections (Belluscio et al., 2002; Lodovichi et al., 2003),

suggesting tight intrabulbar connections between the medial

and lateral maps. Confirming the precision of the intrabulbar

connections, we observed small axonal patches in the granule

cell layer underlying the medial isofunctional glomeruli in the

ipsilateral MOB when tracers were injected into the lateral M71

or mI7/M71 glomeruli (Figures S1B and S5D). We next tested

whether the medial and lateral maps were represented by

same maps in the AONpE. If the isofunctional glomeruli within

the MOB converge into same areas within the AONpE, then small

tracer injection into the AONpE will retrogradely label two mirror-

symmetric clusters of mitral/tufted cells in the ipsilateral MOB.

Small injections into the AONpE retrogradely labeled only a single

cluster of mitral/tufted cells instead of two mirror-symmetric

clusters in the ipsilateral MOB (Figures 3D–3F; Figure S6; n = 9

mice). Consistently, tracer injections into the medial and lateral

glomeruli labeled distinct clusters of axonal terminals in the

AONpE (Figure 1; Figure S3). These results strongly suggest

that the two mirror-symmetric olfactory maps within the MOB

are represented by spatially distinct areas of the AONpE.

Our data indicate that mitral/tufted cells of a defined glomeru-

lus project topographically to a small region within the ipsilateral

AONpE. In turn, neurons in this AONpE region project to the

contralateral granule cell layer directly beneath the bilaterally

symmetric glomerulus associated with the same odorant recep-

tor. The topographic relationship between the MOB and the

AONpE is illustrated in Figure 3G. Linking of bilateral isofunc-

tional olfactory columns by the AONpE is depicted schematically

in Figure 3H.

Contralateral Relay of Specific Olfactory SensoryInformation Requires the AONpEThe linking of isofunctional olfactory columns in bilateral olfac-

tory bulbs suggested that odorant-specific information might

be relayed across bilateral bulbs by the AONpE. We therefore

tested whether odor-evoked activity within one MOB could be

transferred to the contralateral bulb, and if so, whether such

contralateral activation required the AONpE. Following an estab-

lished stimulation protocol (Schaefer et al., 2001), we used c-Fos

as a marker of odor-evoked neuronal activity within the MOB. We

found that acetophenone (PMK), a known ligand for M71 recep-

tor (Bozza et al., 2002), evoked c-Fos expression in both bulbs

when presented to mice with both nares open. PMK robustly ac-

tivated periglomerular cells and granule cells associated with

each M71 glomerulus and adjacent glomeruli (Figures 4A–4D;

n = 6). When PMK was applied while a single nostril was oc-

cluded, c-Fos expression was absent from the glomerular layer

of the occluded side (Figures 4E–4H). However, c-Fos labeling

was observed in a substantial number of cells within the granule

cell layer of the occluded side, although the overall density of

c-Fos immunopositive cells was significantly reduced (Figures

4E and 4F; 60.5 ± 10.0 cells/0.04 mm2 for both nares open versus

17.8 ± 3.1 cells/0.04 mm2 for ipsilateral naris occluded; mean ±

SD; n = 6 mice, p < 0.001, between-group t test). Activated gran-

ule cells on the occluded side were clustered into areas beneath

the M71 glomerulus and adjacent glomeruli, and formed an acti-

vation map symmetric to that seen in the bulb ipsilateral to the

open naris (Figure 4E). The activation of granule cells within bilat-

erally symmetric regions in contralateral MOB strongly sug-

gested odorant-driven activation of contralateral granule cells

instead of nonselective activation by enhanced mechanosen-

sory input following naris occlusion (Grosmaitre et al., 2007).

We confirmed that granule cells can be activated by contralateral

glomerular activity by electrically stimulating the lateral M71 glo-

merulus. Consistent with odorant-driven c-Fos expression, elec-

trical stimulation activated granule cells below the lateral M71

glomerulus in the contralateral bulb in addition to periglomerular

(H) A schematic diagram summarizing our data on the connectivity patterns linking the MOB to the ipsilateral AONpE and then to the contralateral MOB. AC,

anterior commissure; EPL, external plexiform layer.

Scale bars, 200 mm (B, E, and F); 100 mm (C).


Neuron


Figure 4. Odorants Passing through One

Naris Activate Isofunctional Olfactory

Columns in the Contralateral MOB

(A) Drawing representing one highly symmetric

activation map as indicated by c-Fos immunore-

activity following application of acetophenone

(PMK). Green spots indicate the M71 glomeruli;

blue dots, c-Fos+ cells. Coronal section contain-

ing the medial M71 glomeruli is shown here to bet-

ter illustrate the symmetry.

(B) Original c-Fos immunoreactivity of region

shown in the dashed box in (A).

(C and D) c-Fos+ periglomerular cells surrounding

the M71 glomeruli (labels and arrows in [B]).

(E–H) A substantial number of granule cells in the

MOB of the occluded side were c-Fos+ following

PMK application to the contraocclusion naris. (E)

c-Fos expression patterns in the bilateral bulbs.

The right-hand naris was occluded in this case.

(F) c-Fos expression pattern in the granule cell

layer beneath the M71 glomerulus of the occluded

side. (G) c-Fos+ periglomerular cells surrounding

the M71 glomerulus of the open side (left arrow

in [E]). (H) Lack of c-Fos immunoreactivity in the

glomerular layer of the occluded side (right arrow

in [E]) indicates effective occlusion.

(I–K) Drawing of c-Fos+ cells following electrical

stimulation of M71 glomerulus on dorsal MOB (I).

This stimulation activated neurons surrounding

the stimulated glomerulus (J) and the granule cells

under the M71 glomerulus in the contralateral

MOB (K).

Scale bars, 200 mm (B and F); 100 mm (J and K).

cells surrounding the M71 glomerulus ipsilateral to stimulation

(Figures 4I–4K; n = 3 mice).

Physiological recordings were carried out to further examine

the effect of neuronal activation by olfactory responses of the

contralateral MOB. Bulbar neurons were recorded extracellularly

from urethane-anaesthetized mice. Neuronal responses to odor-

ant pulses were recorded and compared before and after contra-

lateral naris occlusion. Naris occlusion was monitored by local

field potentials recorded from the glomerular layer in the bulb

ipsilateral to the occluded naris (Figure 5A, top traces). Putative

granule cells were selected based on low spontaneous activity,

a bursting firing pattern, and recording sites 100–500 mm ventral

to the dorsal mitral cell layer. In contrast, mitral/tufted cells were

selected based on high spontaneous firing coupled to respiratory

rhythm and recording sites within a thin layer 200–300 mm ventral

to the dorsal surface. Among 29 granule cells that showed strong

olfactory responses to odorants, a substantial number of them

(n = 14/29 or �48%) exhibited significant change in odorant-

evoked responses following contralateral naris occlusion

(p < 0.05, t test). Of the 14 cells with significant changes, a vast

majority showed reversible reduction in odor-evoked responses

following contralateral naris occlusion (Figure 5A; change of nor-

malized response intensity = �54.7% ± 20.8%; n = 12/14 cells),


although most of these cells remained responsive to ipsilateral

odorant application. A substantial number of putative mitral/

tufted cells (n = 12/25) showed significant changes in their olfac-

tory responses following contralateral naris occlusion as well.

Five of these twelve mitral/tufted cells with significant changes

exhibited a reversible increase in response strength (Figure 5B;

change in response intensity = 70.0% ± 48.9%), whereas the re-

maining seven cells exhibited reversible decreases (Figure 5C;

change in response intensity =�81.3% ± 45.9%; n = 7/12 cells).

Interestingly, we also observed significant changes of spontane-

ous firing rates from a substantial number of mitral/tufted cells

(Figure 5C; n = 8/25 cells). Most of these cells (n = 7/8) showed

a reversible increase of spontaneous firing following contralateral

naris occlusion (8.0 ± 5.3 Hz versus 16.7 ± 4.0 Hz, before and

after naris occlusion, respectively; p < 0.001, within group t

test), suggesting that the contralateral input may regulate both

olfactory responses and spontaneous activity of mitral/tufted

cells.

We then tested if the AONpE played a role in activating granule

cells in the contralateral MOB. Bipolar stimulating electrodes

were targeted to the soma layer of the dorsal AONpE of anaes-

thetized mice (Figure 6A). Following local electrical stimulation

of the dorsal AONpE, significantly more granule cells were

Neuron


Figure 5. Neuronal Olfactory Responses in the Olfactory Bulb Are Modulated by Contralateral Input

(A) A granule cell’s responses to 0.4% valeraldehyde (left) were dramatically reduced following the occlusion of contralateral naris (middle), and the occlusion

effect was reversible (right). Top traces are the local field potentials recorded from the glomerular layer of the contralateral MOB to confirm the occlusion effect.

Bottom histograms are the peristimulus time histogram of firing rate (spikes/bin) averaged over six to ten trials. In this and following panels, bin width was

determined by the mean duration of the animal’s respiratory cycles.

(B) A mitral cell showed a reversible increase in its response to (-)-isopulegol following the occlusion of contralateral naris. Dashed lines and boxes point to zoom-

in views of spiking activity within the second respiratory cycle during odorant delivery.

(C) A mitral cell showed a reversible decrease in its response strength to propyl acetate and a concomitant increase in spontaneous activity following the occlu-

sion of the contralateral naris.

Horizontal bars under the physiological traces indicate 2 s odorant pulses.

c-Fos+ in the dorsal contralateral MOB compared with those of

unstimulated controls (Figures 6B–6D; 12.3 ± 3.0 cells/0.04 mm2

versus 0.6 ± 1.0 cells/0.04 mm2, with versus without stimulation,

respectively; n = 4 mice, p < 0.001, t test, n = 4 mice). To test

whether the AONpE is necessary for the contralateral activation,

we made small unilateral lesions of the AONpE by pressure-

injecting ibotenic acid (Figure 6E), which kills somata in an area

largely restricted to the AONpE but leaves passing fibers intact

(Figure S7). We then occluded the naris contralateral to the

AONpE lesions and examined c-Fos expression patterns in the

bulbs following the presentation of PMK (n = 6 mice). Virtually

no c-Fos expression was observed in the granule cell layer of

the MOB ipsilateral to the occluded naris and contralateral to

the lesioned AONpE (Figures 6F–6I; 1.2 ± 1.8 cells/0.04 mm2

for occluded side following AONpE lesion; n = 6 mice,

p < 0.001, between-group t test for ipsilateral naris occlusion

with and without AONpE lesion), but the pattern of c-Fos expres-

sion on the side of the open naris appeared normal (Figures 6F

and 6I). Thus, activation of the contralateral MOB requires the

AONpE. In accordance with the role of AONpE in contralateral

transfer of olfactory information, application of PMK activated

ipsilateral AONpE regions comparable to the target area of

M71-associated mitral/tufted cells from mice with contralateral

naris occluded (Figures 6J and 6K).

The AONpE Contributes to InterhemisphericTransfer of Olfactory MemoryOur tracing results showed that the AONpE links bilateral iso-

functional olfactory columns. Our results of c-Fos mapping and

physiological recordings indicated that odorant-specific infor-

mation is relayed between the MOB hemispheres by the AONpE.

Since studies have shown that olfactory memory stored in one

hemisphere can be transferred to the contralateral hemisphere

via the anterior commissure in adult rats (Kucharski and Hall,

1987, 1988), we asked whether the intrabulbar association sys-

tem could contribute to transferring olfactory information that

guides behavioral learning. An adult mouse was allowed to

explore freely in two opposing arms of a T-maze and one of

two odorants (PMK and amyl acetate [AA], 0.5% saturated

vapor) was randomly infused into the arm where the mouse


Neuron


settled (Figure 7A). If the odorant was AA, the animal received

a mild footshock in the same arm after a 10 s delay, whereas

PMK was associated with voltage application to a bottom plate

in the opposite arm. After�100 trials, mice learned to avoid foot-

shock by rapidly moving to the opposite arm following AA appli-

cation but staying in the same arm following PMK, with a perfor-

mance asymptote over 95% correct (Figure 7B). Their responses

to odorant cues were tested without footshock the following day.

After establishing this behavioral paradigm of olfactory learn-

ing, we trained mice with one naris open and the other occluded.

Mice were divided into four groups based on whether their

AONpE was intact or bilaterally lesioned before training and

whether they were tested with the training naris or the previously

occluded naris (Figure 7C; n = 4, 4, 5, and 5 mice for Groups 1–4,

respectively). By mixing the drug ibotenic acid with dextran

tracers, we confirmed that most of the AONpE neurons were

killed, although in some cases the toxin appeared to spread to

small areas adjacent to the AONpE (Figure 7E). Mice with AONpE

lesions could be trained to associate odorants with footshock,

Figure 6. The AONpE Is Sufficient and Nec-

essary for Contralateral Transfer of Activity

Pattern in the MOB

(A–D) Electrical stimulation of dorsal AONpE (A)

activated granule cells within the dorsal area of

the contralateral MOB. (B) A drawing of c-Fos+

cells following electrical stimulation. (C and D)

c-Fos immunoreactivity within the ipsilateral MOB

(C) and the contralateral MOB (D). Arrow in (A)

points to the stimulation site visualized with dex-

tran dye near the tip of the stimulation electrode.

(E–H) Contralateral AONpE lesion abolished c-Fos

immunoreactivity in the granule cell layer following

PMK application to the contralateral naris. (E) Le-

sion site within the AONpE soma layer (dashed

line) as visualized by fluorescent tracers mixed

with ibotenic acid. (F) c-Fos activity was apparently

normal in the left bulb but was eliminated in the

right bulb following lesion of the left AONpE by ibo-

tenic acid and occlusion of the right naris. (G and H)

c-Fos immunoreactivity within the left (G) and right

(H) dashed boxes in (F).

(I) Group data showing the density of the c-Fos+

granule cells beneath M71 glomeruli following

blank control (no odorant), occlusion of ipsilateral

naris and AONpE lesion (ipsi 3 lesion), occlusion

of ipsilateral naris with AONpE intact (ipsi 3), and

PMK application only. Error bars = SD.

(J and K) PMK-evoked c-Fos immunoreactivity

was mostly limited to the dorsal AONpE region.

(K) shows the c-Fos+ neurons within the dashed

box in (J).

Scale bars, 100 mm.

indicating that the AONpE lesion did not

result in general anosmia. When tested

with the training naris, mice performed

equally well with the AONpE either intact

or lesioned (Figure 7D; 97.5% ± 2.5%

and 95.0% ± 3.9% at asymptote for

Groups 1 and 3, respectively; mean ±

SEM for all behavioral assays; p = 0.63, t test), suggesting that

olfactory learning within the ipsilateral hemisphere was indepen-

dent of the AONpE. When tested with the previously occluded

naris, mice performed similarly well with a sham-lesioned

AONpE (Figure 7D; correct ratio = 91.0% ± 5.1%; p = 0.58,

between-group t test for Groups 1 and 2). However, their perfor-

mance dropped to near-chance level when tested with the pre-

viously occluded naris and lesions of the AONpE (Figure 7D; cor-

rect ratio = 60.8% ± 5.0%, p < 0.01, t test for Groups 2 and 4).

Thus, these results are consistent with the concept that the

AONpE contributes to bilateral exchange of information about

odorant identity.

DISCUSSION

In this study, we have integrated tract tracing in transgenic and

wild-type mice, c-Fos labeling, electrophysiology, and behav-

ioral assays to examine the connectivity and functional role of

the interbulbar association system. Our use of transgenic mouse


Neuron


lines with GFP-labeled glomeruli permitted a level of mapping

that has been impossible in previous studies. By depositing tract

tracers focally into single identified glomeruli, we find that mitral/

tufted cells associated with a specific glomerulus project topo-

graphically to the AONpE, where the AONpE neurons in turn pro-

ject to granule cells underlying the isofunctional glomerulus in the

contralateral MOB. Previous studies observed no topography in

the longitudinal direction and quadrant-to-quadrant topography

in other directions for the interbulbar association system (Davis

and Macrides, 1981; Haberly and Price, 1978; Schoenfeld and

Macrides, 1984; Scott et al., 1985). This conflict is likely a result

of large tracer injection sites in previous studies. Furthermore,

these early studies were carried out prior to the discovery of

odorant receptors and their symmetric representation in the bi-

lateral olfactory bulbs. Because of the lack of receptor identity

for the bilaterally isofunctional glomeruli, it had remained unclear

what the interbulbar association system connected exactly. By

targeted tract tracing, we have uncovered a link between iso-

functional olfactory columns in the bilateral olfactory bulbs by

the interbulbar association system. More importantly, as sug-

Figure 7. Interhemispheric Information Exchange in the

Olfactory System Depends on the AONpE

(A) Method of testing olfactory conditioning in a T-maze. Ten sec-

onds after initiation, amyl acetate (AA) application was paired with

footshock in the same arm and PMK with shock in the opposite

arm. Mice learned to move to the opposite arm only after AA

application.

(B) A raw trace showing a mouse’s movement after training. Note

that the mouse quickly moved to the opposite arm following AA

(curved arrows) whereas it stayed in the same arm following

PMK. Vertical arrows indicate valve opening for odorant applica-

tion.

(C) Paradigms for testing the role of the AONpE in contralateral

memory transfer. Mice were trained with only one naris open

and then tested with either the training naris (Groups 1 and 3) or

the previously occluded naris (Groups 2 and 4) with the AONpE

either intact (Groups 1 and 2) or lesioned (Groups 3 and 4). In

Group2, mice were sham-lesioned by pressure injection of vehicle.

(D) Test scores for Groups 1–4 shown in (C). Error bars = SEM.

(E) A representative lesion site in the AONpE. Dashed line indicates

soma layer of the AONpE as revealed by DAPI labeling. Scale bar,

200 mm.

gested by the exquisite anatomical linkage, we find

that the AONpE mediates bilateral exchange of odor-

ant-specific signals.

Consistent with the precision of integration within

the AONpE, morphological studies have shown that

AONpE pyramidal neurons possess small dendritic

trees (Reyher et al., 1988; Scott et al., 1985), a special-

ization that could permit maintenance of odorant

selectivity originating in the MOB. Future studies on

the synaptic distribution between bulbar neurons

and the dendritic tree of AONpE neurons will allow

precise quantification of the topography and conver-

gence ratio. Nevertheless, the highly topographic

and stereotypical representation of each glomerulus

in the AONpE is reminiscent of the projection neurons

in Drosophila antennal lobes, an area comparable to the mam-

malian olfactory bulb. These projection neurons possess stereo-

typed patterns of axonal distribution in the protocerebrum (Jeffe-

ris et al., 2007; Marin et al., 2002; Wong et al., 2002).

Genetic studies have contributed substantially to our under-

standing of the molecular mechanisms underlying the wiring in

the MOB (Wang et al., 1998; Yoshihara et al., 1997; Yu and

Bargmann, 2001) and the Drosophila antennal lobe (Komiyama

et al., 2007). It will be interesting to examine the developmental

mechanisms underlying the precise linking of�1800 pairs of glo-

meruli by the interbulbar association system. The topographic

precision of the interbulbar association system is comparable

to that of the intrabulbar association system, which precisely

connects the two mirror-symmetric olfactory maps within the

MOB (Belluscio et al., 2002; Liu and Shipley, 1994; Lodovichi

et al., 2003; Schoenfeld et al., 1985). These two intrabulbar mir-

ror-symmetric maps appear to be represented by distinct areas

in the AONpE, suggesting that both receptor identity and glomer-

ular spatial location interact to establish the linking of bilateral

olfactory bulbs. The refinement of the intrabulbar connections


Neuron


is activity dependent (Marks et al., 2006), a subject that remains

to be explored in the interbulbar association system.

Olfactory memory formed in one hemisphere can be trans-

ferred to the contralateral hemisphere in rats and humans

(Kucharski and Hall, 1987, 1988; Mainland et al., 2002). The

specific neural circuitry underlying this contralateral transfer of

olfactory memory was unknown. Our c-Fos mapping and phys-

iological recordings show that the activity of specific glomeruli

within one hemisphere of the MOB is transferred to the granule

cells underlying the isofunctional glomeruli in the contralateral

hemisphere. In behavioral assays, mice showed clear deficits

in contralateral transfer of olfactory memory when their AONpE

was lesioned with ibotenic acid. Thus, our experiments suggest

that the AONpE is a critical locus for bilateral relay of olfactory

signals, although cortical areas other than the AONpE can also

contribute to this process. We hypothesize the following sce-

nario in olfactory signal propagation between hemispheres: the

activation of a specific glomerulus activates its associated

mitral/tufted cells, which in turn activate their postsynaptic target

neurons in the AONpE, followed in turn by their targets: the gran-

ule cells underlying the isofunctional glomeruli in the contralat-

eral MOB. The activity of these granule cells may in turn enhance

the activity of mitral/tufted cells associated with the isofunctional

glomeruli by mechanisms such as lateral inhibition (Luo and

Katz, 2001; Yokoi et al., 1995). This scenario may be directly

tested by targeted recordings from mitral/tufted cells associated

with a specific glomerulus and optical stimulation of the contra-

lateral isofunctional glomerulus using a recently developed

transgenic mouse line expressing Channelrhodopsin-2 in

mitral/tufted cells (Arenkiel et al., 2007).

The interbulbar association system may be important for

other behaviors as well. Rats can localize olfactory pulses in

space as arriving from either the left or the right side (Rajan

et al., 2006). The linking of isofunctional columns between the

bulbs may contribute to ‘‘stereo’’ olfactory perception by allow-

ing the system to calculate the bilateral differences of odor-

evoked activity patterns between the isofunctional columns. In

addition, human and animal nasal passages exhibit spontane-

ous change in unilateral airflow resistance (nasal-cycle conges-

tion) (Stoksted, 1953). The AONpE may balance the activity of

the bilateral olfactory bulbs when the same odorant generates

sensory signals of distinct intensities in the bilateral olfactory

epithelia during different phases of nasal cycles (Sobel et al.,

1999).

The two symmetric hemispheres of mammalian forebrain are

connected by dense fibers coursing within the corpus callosum

and anterior commissure. The studies on ‘‘split-brains’’ illustrate

the essential role of interhemispheric communication in percep-

tion and sensory learning (Gazzaniga, 1995). For example, ani-

mals with severed corpus callosum have bilaterally separate

‘‘learning centers,’’ with one hemisphere failing to access the

learned information in another hemisphere (Myers and Sperry,

1958). Our use of transgenic mice with a specific glomerulus

expressing GFP permitted unprecedented mapping of bilateral

connections for specific olfactory glomeruli. These data indicate

a highly structured organization of anatomical pathways carrying

specific olfactory information between the bilateral bulbs, sug-

gesting the interbulbar association system as an advantageous


model for studying the cellular circuits, function, and develop-

ment of interhemispheric connections.

EXPERIMENTAL PROCEDURES

Adult M71-IRES-tauGFP and mI7/M71-IRES-tauGFP mice (7–12 weeks old)

were used in all experiments except for T-maze assays, in which C57BL/6

mice were used.

Tract Tracing and Ibotenic Acid Lesion

Mice were anaesthetized with a mixture of ketamine (200 mg/kg) and xylazine

(25 mg/kg, i.p.). Guided by green fluorescence of the M71 or mI7/M71 glo-

merulus, biotinylated dextran amines (10K MW, BDA, 10%, Invitrogen) mixed

with Texas red dextran amines (10K MW, TRDA, 1%) or fluorescein-

conjugated dextran amines (10K MW, FDA, 10%) were iontophoretically

injected (1 mA, 1–20 s) into target areas through a micropipette. For lesion of

the AONpE, 0.1–0.5 ml of 0.5% ibotenic acid in saline was pressure injected

into the AONpE. Ibotenic acid was mixed with 1% tetramethylrhodamine

dextran amine (TMR-DA) to visualize the lesion sites. Sham lesions were

carried out by pressure injection of vehicle (1% TMR-DA in saline).

Histology

Mice were overdosed with pentobarbital (300 mg/kg, i.p.) and perfused trans-

cardially with 0.1M phosphate buffered saline (PBS) and then 4% paraformal-

dehyde in PBS. Following postfixation and cryoprotection, coronal sections

(60 mm) were prepared on a freezing microtome (Leica CR 1900). BDA tracers

were visualized by fluorescence-conjugated streptavidin. Sections were

mounted in 50% glycerol with DAPI.

Mapping Neuronal Activity with c-Fos

For odorant exposure, mice were placed overnight in a chamber (volume = 1.8

liters) with continuous airflow (4 l/min) filtered with active charcoal. PMK pulses

(�1% saturated vapor) were presented for 2 min at 5 min intervals over

a 15 min period. For electrical stimulation, mice were first anaesthetized with

urethane (1.64 g/kg injected as 20% solution i.p.). After 4 hr of exposure to

purified air, bipolar tungsten electrodes (1 MU, Microprobes) were placed

within the M71 glomerulus or the AONpE to generate 240 trains of electric

pulses over 20 min, with each train comprising 100 pulses at 30–50 mA

(100 Hz; pulse duration = 0.1 ms). Mice were transcardially perfused 90 min af-

ter either odorant or electrical stimulation and brains were later processed for

c-Fos immunostaining. The primary antibody was a rabbit polyclonal antibody

against c-Fos (1:500, Santa Cruz) and the secondary antibody was Cy3-

conjugated goat anti-rabbit (1:500, Jackson ImmunoResearch).

Confocal Imaging and Analysis

Sections were imaged with a laser scanning confocal microscope (Zeiss LSM

510). Optical stacks (Z-interval 1.0 mm) were projected into single frames.

Whole-view images were constructed by assembling multiple adjacent frames

of 103 images. Borders and cell layers were identified by DAPI labeling and

autofluorescence. To define c-Fos+ cells, we first measured the mean and

SD of background optical signals. Cells with optical intensities >[mean +

3*SD] were considered c-Fos+. The density of c-Fos+ neurons was calculated

by counting the number of c-Fos+ neurons within a square with side length

equal to 200 mm, which is roughly the size of large glomeruli in the olfactory

bulb.

Electrophysiological Recordings

Animals were anaesthetized with urethane. After the dorsal M71 glomerulus

was identified, a small piece of bone above the glomerulus was removed

and tungsten electrodes were lowered below the M71 glomerulus. Odorants

(1% saturated vapor, 2 s pulse duration) were applied with a custom-made

64-channel robotic olfactometer. Data were band-pass filtered at 0.5–

3.0 KHz, digitized at 6 KHz, and analyzed with a custom-written Matlab pro-

gram. For a neuron to be considered responsive to an odorant, its mean firing

rate during odorant application needed to be significantly different (p < 0.05,

paired t test) from its mean spontaneous rate within 2 s preceding stimulus

Neuron


onset. Its response intensity to an odorant was calculated by subtracting the

spontaneous rate from the firing rate during the odorant application. Change

of normalized response intensity was calculated as ([response intensity after

naris occlusion] / [response intensity before naris occlusion]) � 100%.

Behavioral Assay

A detailed description of this method can be found in Supplementary Online

Material. Briefly, adult C57BL/6 mice were trained in a customized T-maze

to associate AA with footshock in the same arm and PMK with footshock in

the opposite arm. Mice learned to avoid footshock by moving to the opposite

arm following AA or staying within the same arm following PMK. After learning

(100% correct for ten trials), they were tested without actual application of volt-

age. A nostril was occluded by inserting a soft rod (0.8 mm in diameter, 4 mm in

length) and then sealing with a small drop of Vetbond tissue adhesive. Animal

movement was monitored with an infrared CCD camera. Training and testing

were controlled by a custom-written Matlab program.

SUPPLEMENTAL DATA

The Supplemental Data for this article can be found online at http://www.

neuron.org/cgi/content/full/58/4/613/DC1/.

ACKNOWLEDGMENTS

We thank Peter Mombaerts and Stuart Firestein for M71-IRES-tauGFP and

mI7/M71-IRES-tauGFP mice; Liqun Luo, Abigail Person, Kazunari Miyami-

chi, and Charles Zuker for comments on the manuscript; and Ji Hu for assis-

tance with Matlab programming. This work was supported by China Ministry

of Science and Technology 863 Projects, Human Frontier Science Program,

and National Natural Science Foundation of China.

Received: October 5, 2007

Revised: January 24, 2008

Accepted: March 8, 2008

Published: May 21, 2008

REFERENCES

Arenkiel, B.R., Peca, J., Davison, I.G., Feliciano, C., Deisseroth, K., Augustine,

G.J., Ehlers, M.D., and Feng, G. (2007). In vivo light-induced activation of neu-

ral circuitry in transgenic mice expressing channelrhodopsin-2. Neuron 54,

205–218.

Belluscio, L., Lodovichi, C., Feinstein, P., Mombaerts, P., and Katz, L.C.

(2002). Odorant receptors instruct functional circuitry in the mouse olfactory

bulb. Nature 419, 296–300.

Bozza, T., Feinstein, P., Zheng, C., and Mombaerts, P. (2002). Odorant recep-

tor expression defines functional units in the mouse olfactory system.

J. Neurosci. 22, 3033–3043.

Buck, L., and Axel, R. (1991). A novel multigene family may encode odorant

receptors: a molecular basis for odor recognition. Cell 65, 175–187.

Davis, B.J., and Macrides, F. (1981). The organization of centrifugal projections

from the anterior olfactory nucleus, ventral hippocampal rudiment, and piri-

form cortex to the main olfactory bulb in the hamster: an autoradiographic

study. J. Comp. Neurol. 203, 475–493.

Gazzaniga, M.S. (1995). Principles of human brain organization derived from

split-brain studies. Neuron 14, 217–228.

Grosmaitre, X., Santarelli, L.C., Tan, J., Luo, M., and Ma, M. (2007). Dual func-

tions of mammalian olfactory sensory neurons as odor detectors and mechan-

ical sensors. Nat. Neurosci. 10, 348–354.

Haberly, L.B., and Price, J.L. (1978). Association and commissural fiber sys-

tems of the olfactory cortex of the rat. J. Comp. Neurol. 178, 711–740.

Jefferis, G.S., Potter, C.J., Chan, A.M., Marin, E.C., Rohlfing, T., Maurer, C.R.,

Jr., and Luo, L. (2007). Comprehensive maps of Drosophila higher olfactory

centers: spatially segregated fruit and pheromone representation. Cell 128,

1187–1203.

Komiyama, T., Sweeney, L.B., Schuldiner, O., Garcia, K.C., and Luo, L. (2007).

Graded expression of semaphorin-1a cell-autonomously directs dendritic tar-

geting of olfactory projection neurons. Cell 128, 399–410.

Kucharski, D., and Hall, W.G. (1987). New routes to early memories. Science

238, 786–788.

Kucharski, D., and Hall, W.G. (1988). Developmental change in the access to

olfactory memories. Behav. Neurosci. 102, 340–348.

Liu, W.L., and Shipley, M.T. (1994). Intrabulbar associational system in the rat

olfactory bulb comprises cholecystokinin-containing tufted cells that synapse

onto the dendrites of GABAergic granule cells. J. Comp. Neurol. 346, 541–558.

Lodovichi, C., Belluscio, L., and Katz, L.C. (2003). Functional topography of

connections linking mirror-symmetric maps in the mouse olfactory bulb.

Neuron 38, 265–276.

Luo, M., and Katz, L.C. (2001). Response correlation maps of neurons in the

mammalian olfactory bulb. Neuron 32, 1165–1179.

Luo, L., and Flanagan, J.G. (2007). Development of continuous and discrete

neural maps. Neuron 56, 284–300.

Mainland, J.D., Bremner, E.A., Young, N., Johnson, B.N., Khan, R.M., Bensafi,

M., and Sobel, N. (2002). Olfactory plasticity: one nostril knows what the other

learns. Nature 419, 802.

Marin, E.C., Jefferis, G.S., Komiyama, T., Zhu, H., and Luo, L. (2002). Repre-

sentation of the glomerular olfactory map in the Drosophila brain. Cell 109,

243–255.

Marks, C.A., Cheng, K., Cummings, D.M., and Belluscio, L. (2006). Activity-

dependent plasticity in the olfactory intrabulbar map. J. Neurosci. 26,

11257–11266.

Mombaerts, P., Wang, F., Dulac, C., Chao, S.K., Nemes, A., Mendelsohn, M.,

Edmondson, J., and Axel, R. (1996). Visualizing an olfactory sensory map. Cell

87, 675–686.

Myers, R.E., and Sperry, R.W. (1958). Interhemispheric communication

through the corpus callosum: mnemonic carry-over between the hemispheres.

AMA Arch. Neurol. Psychiatry 80, 298–303.

O’Leary, D.D., Yates, P.A., and McLaughlin, T. (1999). Molecular development

of sensory maps: representing sights and smells in the brain. Cell 96, 255–269.

Rajan, R., Clement, J.P., and Bhalla, U.S. (2006). Rats smell in stereo. Science

311, 666–670.

Ressler, K.J., Sullivan, S.L., and Buck, L.B. (1994). Information coding in the

olfactory system: evidence for a stereotyped and highly organized epitope

map in the olfactory bulb. Cell 79, 1245–1255.

Reyher, C.K., Schwerdtfeger, W.K., and Baumgarten, H.G. (1988). Interbulbar

axonal collateralization and morphology of anterior olfactory nucleus neurons

in the rat. Brain Res. Bull. 20, 549–566.

Rubin, B.D., and Katz, L.C. (1999). Optical imaging of odorant representations

in the mammalian olfactory bulb. Neuron 23, 499–511.

Schaefer, M.L., Young, D.A., and Restrepo, D. (2001). Olfactory fingerprints for

major histocompatibility complex-determined body odors. J. Neurosci. 21,

2481–2487.

Schoenfeld, T.A., and Macrides, F. (1984). Topographic organization of con-

nections between the main olfactory bulb and pars externa of the anterior

olfactory nucleus in the hamster. J. Comp. Neurol. 227, 121–135.

Schoenfeld, T.A., Marchand, J.E., and Macrides, F. (1985). Topographic orga-

nization of tufted cell axonal projections in the hamster main olfactory bulb: an

intrabulbar associational system. J. Comp. Neurol. 235, 503–518.

Scott, J.W., Ranier, E.C., Pemberton, J.L., Orona, E., and Mouradian, L.E.

(1985). Pattern of rat olfactory bulb mitral and tufted cell connections to the an-

terior olfactory nucleus pars externa. J. Comp. Neurol. 242, 415–424.

Shepherd, G.M., Chen, W.R., and Greer, C.A. (2004). Olfactory bulb. In The

Synaptic Organization of the Brain, G.M. Shepherd, ed. (New York: Oxford

University Press), pp. 165–216.

Sobel, N., Khan, R.M., Saltman, A., Sullivan, E.V., and Gabrieli, J.D. (1999). The

world smells different to each nostril. Nature 402, 35.


http://www.neuron.org/cgi/content/full/58/4/613/DC1/

http://www.neuron.org/cgi/content/full/58/4/613/DC1/

Neuron


Stoksted, P. (1953). Rhinometric measurements for determination of the nasal

cycle. Acta Otolaryngol. Suppl. 109, 159–175.

Uchida, N., Takahashi, Y.K., Tanifuji, M., and Mori, K. (2000). Odor maps in the

mammalian olfactory bulb: domain organization and odorant structural fea-

tures. Nat. Neurosci. 3, 1035–1043.

Vassar, R., Chao, S.K., Sitcheran, R., Nunez, J.M., Vosshall, L.B., and Axel, R.

(1994). Topographic organization of sensory projections to the olfactory bulb.

Cell 79, 981–991.

Wachowiak, M., and Cohen, L.B. (2001). Representation of odorants by recep-

tor neuron input to the mouse olfactory bulb. Neuron 32, 723–735.

Wang, F., Nemes, A., Mendelsohn, M., and Axel, R. (1998). Odorant receptors

govern the formation of a precise topographic map. Cell 93, 47–60.

Willhite, D.C., Nguyen, K.T., Masurkar, A.V., Greer, C.A., Shepherd, G.M., and

Chen, W.R. (2006). Viral tracing identifies distributed columnar organization in

the olfactory bulb. Proc. Natl. Acad. Sci. USA 103, 12592–12597.


Wong, A.M., Wang, J.W., and Axel, R. (2002). Spatial representation of the glo-

merular map in the Drosophila protocerebrum. Cell 109, 229–241.

Yokoi, M., Mori, K., and Nakanishi, S. (1995). Refinement of odor molecule tun-

ing by dendrodendritic synaptic inhibition in the olfactory bulb. Proc. Natl.

Acad. Sci. USA 92, 3371–3375.

Yoshihara, Y., Kawasaki, M., Tamada, A., Fujita, H., Hayashi, H., Kagamiyama,

H., and Mori, K. (1997). OCAM: A new member of the neural cell adhesion mol-

ecule family related to zone-to-zone projection of olfactory and vomeronasal

axons. J. Neurosci. 17, 5830–5842.

Yu, T.W., and Bargmann, C.I. (2001). Dynamic regulation of axon guidance.

Nat. Neurosci. Suppl. 4, 1169–1176.

Zhang, X., and Firestein, S. (2002). The olfactory receptor gene superfamily of

the mouse. Nat. Neurosci. 5, 124–133.

Neuron

Previews

Interhemispheric Olfactory Circuitand the Memory Beyond

Takeshi Imai1 and Hitoshi Sakano1,*1Department of Biophysics and Biochemistry, Graduate School of Science, The University of Tokyo, Tokyo 113-0032, Japan*Correspondence: [email protected] 10.1016/j.neuron.2008.05.004

In mammals, olfactory sensory neurons project their axons exclusively to the ipsilateral olfactory bulb. Itremains unclear how odor information interacts between the two hemispheres of the brain. In this issue ofNeuron, Yan et al. describe the precise interbulbar connection through the anterior olfactory nucleus parsexterna (AONpE), which links contralateral isotypic olfactory columns.

Since the discovery of odorant receptors

(ORs) (Buck and Axel, 1991), there has

been tremendous progress in under-

standing how odor information is repre-

sented in the olfactory bulb (OB), the first

center for processing odor information.

Airborne odorants are detected with

�1000 types of ORs expressed by olfac-

tory sensory neurons (OSNs) in the olfac-

tory epithelium (OE). Each

OSN expresses only one

functional OR gene in a mono-

allelic manner, following the

one neuron-one receptor rule

(reviewed by Serizawa et al.,

2004). Furthermore, OSNs

expressing a given type of

OR converge their axons to

a specific pair of glomeruli in

each OB, referred to as the

one glomerulus-one receptor

rule (reviewed by Imai and

Sakano, 2007). The result is a

stereotyped, mirror-symmet-

ric glomerular map (Figure 1).

Thus, the odor information

detected in the OE is con-

verted to a topographic map

of activated glomeruli (re-

viewed by Mori et al., 2006).

In the OB, each glomerulus

with its underlying neurons

forms a functional unit called

the ‘‘olfactory column’’ for

odor processing (reviewed

by Wilson and Mainen,

2006). Local interneurons,

such as periglomerular cells

and granule cells, tune the

odor-derived signals that are

conveyed to second-order

neurons, the mitral/tufted

(M/T) cells. Each M/T cell receives direct

inputs from a single glomerulus and sends

branched axons to several distinct re-

gions in the olfactory cortex (OC), as well

as to other brain regions involved in multi-

modal sensory integration. A recent study

demonstrated that activation of specific

sets of glomeruli in the dorsal domain of

the OB elicits stereotyped innate behav-

iors, suggesting the presence of a hard-

wired neural circuit in the mammalian

main olfactory system (Kobayakawa

et al., 2007). In contrast, discrimination

and associative learning of odor informa-

tion appear to be more complex and

plastic. Fundamental questions regarding

these processes remain mostly unan-

swered. It is yet to be elucidated how

odors are recognized and dis-

criminated and how olfactory

memories are stored and re-

trieved. The first step toward

answering these questions

is to describe the neural cir-

cuitry for respective olfactory

responses from both anatom-

ical and functional aspects.

In this issue of Neuron,

Yan et al. (2008) address an

important question for odor

perception: how are the olfac-

tory inputs from two nostrils

coordinated in the brain? In

contrast to retinal projection

in binocular animals, mam-

malian OSNs project their

axons solely to the ipsilateral

OB: right nostril to the right

OB and left nostril to the

left OB. The same is true for

second-order neurons: the

M/T cells project their axons

to the ipsilateral OC. None-

theless, behavioral studies

demonstrate that the olfac-

tory memory learned by one

nostril can be recalled by

stimuli to either nostril (Ku-

charski and Hall, 1987), sug-

gesting the existence of an

interhemispheric connection

Figure 1. Intrabulbar (Lodovichi et al., 2003) and Interbulbar(Yan et al., 2008) Circuitries in the Mouse Olfactory SystemA horizontal section of the mouse olfactory system is schematically shown.Information flow from one particular lateral glomerulus in the left OB is shown.Isofunctional glomeruli are shown in green. Excitatory and inhibitory neuronsare shown in red and blue, respectively. External tufted cells project axonsto the granule cells in the isofunctional olfactory columns, to form intrabulbarconnections (Lodovichi et al., 2003). M/T cells project axons to the OC, includ-ing the AON. In the AONpE, M/T cells synapse to excitatory neurons, which inturn project axons to the granule cells in the isofunctional olfactory column inthe contralateral OB. This disynaptic circuitry forms the interbulbar connec-tion (Yan et al., 2008). AC, anterior commissure; ETC, external tufted cell; GC,granule cell.

Neuron 58, May 22, 2008 ª2008 Elsevier Inc. 465


Neuron

Previews

for olfactory learning and/or recall. Which

neural circuit mediates the interhemi-

spheric communication? And how?

The anterior olfactory nucleus (AON)

is a region of the OC known to contain

neurons that project to the contralateral

OB. This contralateral projection passes

through the anterior commissure, one of

the fiber bundles connecting the two

hemispheres of the brain. Notably, the

anterior olfactory nucleus pars externa

(AONpE) receives inputs from M/T cells

in the ipsilateral OB and sends fibers

exclusively to the granule cell layer in the

contralateral OB. Among the AON subre-

gions, only the AONpE exhibits topo-

graphic projection to the OB (Davis and

Macrides, 1981). In the present study,

Yan et al. (2008) analyzed whether or not

connections through the AONpE link the

OR-specific olfactory columns of oppo-

site OBs. When fluorescent tracers were

injected into a glomerulus for OR-M71,

anterograde labeling was detected in a

specific spot within the ipsilateral AONpE.

Anterograde labeling from various glo-

merular locations revealed a conservation

of topography between the OB and

AONpE: relative locations of glomeruli in

the OB along the anterior-posterior, dor-

sal-medial, and medial-lateral axes are

all preserved in the AONpE. The projec-

tion of AONpE neurons to the contralat-

eral OB also occurs in a topographically

conserved manner: tracer injection to

the granule cell layer resulted in retro-

grade labeling of a specific spot in the

contralateral AONpE region. To examine

the OR-specific neuronal connection

through the AONpE, Yan et al. (2008) in-

jected red tracers into the M71 glomeru-

lus and green tracers into the granule

cell layer beneath the contralateral M71

glomerulus. They found colocalization of

red and green dyes in a restricted area in

the AONpE. Thus, the two projections,

one from OB to ipsilateral AONpE and

the other from AONpE to contralateral

OB, link the two corresponding OR-spe-

cific ‘‘isofunctional’’ olfactory columns

with high precision (Figure 1).

The next question was, does this circuit

regulate the neuronal activity of the con-

tralateral OB? Using c-Fos expression as

a measure of activity, Yan et al. (2008) an-

alyzed the odor-evoked neuronal activity

in the OB. Odorant stimuli induced c-Fos

expression in the periglomerular cells

466 Neuron 58, May 22, 2008 ª2008 Elsevie

and granule cells in the OB. When one

nostril was occluded, c-Fos signals were

dramatically reduced on the occluded

side. Interestingly, however, significant

levels of signal were detected in granule

cells. The activities of the underlying

granule cells were caused by excitatory

inputs from the contralateral AONpE.

When AONpE neurons on the contralat-

eral (open) side were killed by isobutenic

acid, c-Fos signals in granule cells on the

occluded side were not observed. These

results indicate that M/T cells activate

the ipsilateral AONpE neurons, which in

turn lead to the excitation of granule cells

in the contralateral OB. This notion was

also confirmed by electrophysiological

recordings of granule cell activities.

In each OB, mirror-symmetrical maps

are formed in the medial and lateral

halves, and reciprocal projections within

the OB link the isotypic olfactory columns

(Lodovichi et al., 2003). Connectivity of

the interbulbar circuitry appears to be

analogous to the intrabulbar circuitry. It

has been found that external tufted cells

(excitatory neurons) project to granule

cells within the isotypic olfactory column

on the other side of the OB (Figure 1).

However, it is still unknown how the acti-

vated granule cells modulate M/T cells,

and how the intrabulbar association

tunes olfactory signals as a whole. Be-

cause granule cells are GABAergic neu-

rons, both intrabulbar and interbulbar

connections appear to be inhibitory.

These systems might be used to syn-

chronize the odor-evoked neuronal activ-

ity and transmit the robust signals to the

OC. Alternatively, interbulbar reciprocal

connections may enhance the difference

in inputs between the two nostrils, allow-

ing the ‘‘stereo’’ sensation of odors. In-

deed, there is a report demonstrating

that bilateral sniffing is essential to sense

the direction of the odor source in rat

(Rajan et al., 2006).

Yan et al. (2008) assume a different

scenario for the function of interbulbar

connection. They examined the functional

requirement of interbulbar circuitry for

the exchange of olfactory information

between hemispheres. It has been re-

ported that unilateral olfactory learning

can be recalled by stimulating either of

the two nostrils, trained or untrained,

and that transection of the anterior

commissure impairs the memory recall

r Inc.

by the untrained nostril (Kucharski and

Hall, 1987). Yan et al. (2008) specifically

eliminated the AONpE neurons by focal

injection of isobutenic acid and assessed

its impact on bilateral communication.

They used a unilateral olfactory learning/

recall paradigm in which both ANOpE-le-

sioned and sham-lesioned mice were

trained to associate footshock with a par-

ticular odor through a single nostril and

then tested for the olfactory memory

either with the trained or untrained nostril.

Sham-lesioned mice could avoid the odor

with either nostril. Even when the AONpE

was lesioned, the mice still successfully

avoided the associated odor with the

trained nostril, suggesting that olfactory

learning and recall in the ipsilateral brain

was not affected by the surgery. In con-

trast, the AONpE-lesioned mice did not

avoid the odor with the untrained nostril.

Because the AONpE neurons project ex-

clusively to the contralateral OB and not

to other brain regions (Davis and Macr-

ides, 1981), these results indicate that

the projection from AONpE to contralat-

eral OB is essential for interhemispheric

communication in the olfactory learning

and/or recall. This finding is quite amaz-

ing; however, one must be cautious in

interpreting the data of surgical lesions.

In the vicinity of the AONpE, other sub-

regions are closely packed, some of

which also project to the contralateral

hemisphere (Davis and Macrides, 1981).

Therefore, one cannot exclude the possi-

bility that the AONpE lesion may have

also damaged other subregions, which

might instead mediate the learning and/

or memory recall between the two hemi-

spheres. To address these issues, spe-

cific elimination or silencing of AONpE

neurons using genetic methods (e.g., Ko-

bayakawa et al., 2007) will be useful for

future studies.

The present study by Yan et al. (2008)

also raises some important questions. If

the interbulbar circuitry indeed mediates

the learning and/or recall, does the activa-

tion of granule cells in the contralateral OB

elicit the memory? In order to learn or to

recall the olfactory memory in the brain,

regulation of M/T cells appears to be es-

sential, because they are the only projec-

tion neurons in the OB. Yan et al. (2008) in-

deed find that the neuronal activities of

some M/T cells are changed (either ex-

cited or inhibited) by contralateral inputs.

Neuron

Previews

However, it is puzzling that the activation

of the underlying GABAergic granule cells

leads to such changes. Future studies

using genetic methods that label a specific

subset of neurons or stimulate them by

light-gated ion channels will help describe

the local neural circuitry in detail.

Interbulbar circuit formation is an in-

triguing issue in developmental neurobi-

ology. Recent studies have shed light

on the molecular basis of olfactory map

formation in the OB: OR-derived cAMP

signals direct the axonal projection of

OSNs by regulating the gene expression

of axon guidance/sorting molecules (re-

viewed by Imai and Sakano, 2007). Spon-

taneous signaling from ORs, rather than

odor-evoked activity, appears to be im-

portant in mammalian OSN projection,

whereas the intrabulbar projection of

external tufted cells is highly dependent

on neuronal activity (Marks et al., 2006).

Although the intrabulbar connection is

monosynaptic, the interbulbar circuit de-

scribed in the present study is disynaptic,

which may require more complicated

processing during development. The

commissural fibers play an important

Au Naturel

Garrett B. Stanley1,*1Coulter Department of Biomedical Engineerin*Correspondence: [email protected] 10.1016/j.neuron.2008.05.003

Although adaptation is a ubiquitousquences in the natural visual enviroa comprehensive set of in vivo expof adaptation that have been well stenvironment and are predictable froensembles.

Nature does nothing uselessly.

—Aristotle, 384–322 BC

It is this compelling idea that has driven

neuroscientists for decades to ponder the

evolution, development, and function of

the brain in the context of the natural envi-

role in exchanging higher-order informa-

tion between hemispheres, and studies

on the interbulbar circuitry will provide

new insights into the molecular mecha-

nisms of the precise wiring between

hemispheres.

Another issue raised in the present

study is the interhemispheric exchange

of olfactory information. The reported

failure of interhemispheric communica-

tion in the behavioral experiments on the

AONpE-lesioned mice may be due to the

inability to either ‘‘form’’ the memory or

to ‘‘transfer’’ it to the contralateral hemi-

sphere. Alternatively, it may be due to

the deficit of the ‘‘recall’’ of memory

stored in the contralateral hemisphere.

These possibilities are not mutually exclu-

sive and can be dissected and tested with

genetic tools to silence the AON in a re-

versible manner. It is still not well under-

stood where and how olfactory memory

is stored and what kinds of neuronal activ-

ities lead to learning and recall of olfactory

information. The interbulbar circuitry de-

scribed in the present study will continue

to serve as an excellent tool for the study

of olfactory memory.

g, Georgia Institute of Technology, and Emoryh.edu

property of neurons in the early visnment are unknown. In this issue oferiments in the visual thalamus, thaudied with artificial probes capture tm properties of the visual scene that

ronment within which we exist. Simply

put, to understand the brain, we cannot

ignore our surroundings. Although the

key to the mysteries of the endless com-

plexity of the anatomy and function may

indeed lie at the interface between the

individual and the world, scientific explo-

Neuro

REFERENCES

Buck, L., and Axel, R. (1991). Cell 65, 175–187.

Davis, B.J., and Macrides, F. (1981). J. Comp.Neurol. 203, 475–493.

Imai, T., and Sakano, H. (2007). Curr. Opin. Neuro-biol. 17, 507–515.

Kobayakawa, K., Kobayakawa, R., Matsumoto, H.,Oka, Y., Imai, T., Ikawa, M., Okabe, M., Ikeda, T.,Itohara, S., Kikusui, T., et al. (2007). Nature 450,503–508.

Kucharski, D., and Hall, W.G. (1987). Science 238,786–788.

Lodovichi, C., Belluscio, L., and Katz, L.C. (2003).Neuron 38, 265–276.

Marks, C.A., Cheng, K., Cummings, D.M., and Bel-luscio, L. (2006). J. Neurosci. 26, 11257–11266.

Mori, K., Takahashi, Y.K., Igarashi, K.M., andYamaguchi, M. (2006). Physiol. Rev. 86, 409–433.

Rajan, R., Clement, J.P., and Bhalla, U.S. (2006).Science 311, 666–670.

Serizawa, S., Miyamichi, K., and Sakano, H. (2004).Trends Genet. 20, 648–653.

Wilson, R.I., and Mainen, Z.F. (2006). Annu. Rev.Neurosci. 29, 163–201.

Yan, Z., Tan, J., Qin, C., Lu, Y., Ding, C., and Luo, M.(2008). Neuron 58, this issue, 613–624.

University, Atlanta, GA 30332, USA

ual pathway, the functional conse-Neuron, Mante et al. show, throught the basic functional mechanismshe neuronal response in the naturalmay be represented by local neural

ration of the brain from this perspective

is a vexing task. The idea does not lend

itself well to carefully controlled experi-

ments that normally constitute scientific

investigation. Nevertheless, confronting

this issue may help us move from what

can the brain do? to what does the brain

n 58, May 22, 2008 ª2008 Elsevier Inc. 467


precise circuitry links bilaterally symmetric olfactory maps

Documents