nature neurOSCIenCe  VOLUME 15 | NUMBER 8 | AUGUST 2012 1153

a r t I C l e S

The ability to associate locations with particular events is essential for the survival of all animals. It is widely believed that NMDAR-dependent synaptic plasticity contributes to the formation of associa-tive memories1, and hippocampal NMDARs, particularly those in the dorsal CA1, are thought to be important for associative, long-term spatial memory2, although their precise role remains unclear. Indeed, as recently noted, “even though the synaptic plasticity and memory hypothesis is enshrined in most neuroscience text books, this issue is still far from resolved”3. Pharmacological studies with NMDAR antagonists are equivocal4–6, and previous evidence from transgenic mice lacking NMDARs in dorsal CA1 (refs. 7,8) is confounded by the spread of NMDAR deletion to principal neurons in cortex, as shown in recent studies of this transgenic line9–13.

Mice lacking NMDARs selectively in dentate gyrus granule cells, and hence long-term potentiation (LTP) in lateral and medial perforant path synapses, exhibit normal long-term spatial memory14,15. We have now generated mice (Grin1∆DGCA1) in which NMDARs are also lacking in dorsal and, to a lesser extent, ventral hippocampal CA1 pyramidal cells, specifically in adulthood, thus providing a new mouse model for assess-ing the hippocampal NMDAR and memory hypothesis. Here we investi-gated Grin1∆DGCA1 mice on tests of associative long-term spatial memory. Mice were assessed on reference memory versions of the water maze and radial maze tasks, which both depend on dorsal hippocampus16–21.

RESULTSDentate gyrus– and CA1-specific NMDAR deletionWe used a mouse line (Grin1∆DGCA1; Fig. 1a) gene-targeted for loxP-tagged Grin1 alleles and carrying two transgenes, TgLC1 and TgCN12, which enabled doxycycline (dox)-sensitive, Cre-mediated GluN1

ablation in excitatory hippocampal, but not cortical, neurons of adults22,23 by use of a Camk2a promoter fused to a Grin2c silencer element24. Selective hippocampal NMDAR removal required switch-ing off Cre expression by dox during embryogenesis and nursing the pups with mothers taken off dox postnatally. Delayed Cre expression, detected 4 weeks after birth (Fig. 1b,c), eventually reached a peak in all dentate gyrus granule cells, whereas in CA1 a Cre expression gradi-ent formed along the dorso-ventral axis (Fig. 1d and Supplementary Fig. 1h). We assessed specificity of cumulative Cre expression by 5-bromo-4-chloro-3-indolyl-β-d-galactoside (X-gal) staining in double transgenic Cre indicator mice (TgCN12; TgLC1; Gt(ROSA)26Sor), which demonstrated strong Cre-induced β-galactosidase activity throughout the entire dentate gyrus and mossy fibers, as well as dorsal and, to a lesser extent, ventral CA1 (Fig. 1b,c and Supplementary Fig. 1d–g). All other parts of the hippocampal formation (CA3, lateral entorhinal cortex, medial entorhinal cortex, subiculum) exhibited neg-ligible recombination (<1% co-labeling; Supplementary Fig. 1d–g), although β-galactosidase was observed in olfactory bulb granule cells and approximately 30% of layer II piriform cortex neurons (Fig. 1b,c and Supplementary Fig. 1d).

We confirmed efficient loss of Grin1 expression in dorsal CA1 and dentate gyrus by in situ hybridization, which also revealed reduced expression of the GluA1 AMPA receptor subunit gene Gria1 in the dentate gyrus granule cell layer (Fig. 1e and Supplementary Fig. 2). Further analysis indicated that in older mice the volume of the den-tate gyrus was reduced (~50% at age >1 year). NeuN, calbindin, glial fibrillary acidic protein (GFAP) and parvalbumin expression patterns showed that the upper blade of the dentate gyrus granule cell layer was more compromised than the lower blade, exhibiting

1Department of Experimental Psychology, University of Oxford, Oxford, UK. 2Max Planck Institute for Medical Research, Heidelberg, Germany. 3Department of Physiology, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway. Correspondence should be addressed to D.M.B. (david.bannerman@psy.ox.ac.uk) or P.H.S. (peter.seeburg@mpimf-heidelberg.mpg.de).

Received 21 February; accepted 25 June; published online 15 July 2012; doi:10.1038/nn.3166

Dissecting spatial knowledge from spatial choice by hippocampal NMDA receptor deletionDavid M Bannerman1, Thorsten Bus2, Amy Taylor1, David J Sanderson1, Inna Schwarz2, Vidar Jensen3, Øivind Hvalby3, J Nicholas P Rawlins1, Peter H Seeburg2 & Rolf Sprengel2

Hippocampal NMDA receptors (NMDARs) and NMDAR-dependent synaptic plasticity are widely considered crucial substrates of long-term spatial memory, although their precise role remains uncertain. Here we show that Grin1∆DGCA1 mice, lacking GluN1 and hence NMDARs in all dentate gyrus and dorsal CA1 principal cells, acquired the spatial reference memory water maze task as well as controls, despite impairments on the spatial reference memory radial maze task. When we ran a spatial discrimination water maze task using two visually identical beacons, Grin1∆DGCA1 mice were impaired at using spatial information to inhibit selecting the decoy beacon, despite knowing the platform’s actual spatial location. This failure could suffice to impair radial maze performance despite spatial memory itself being normal. Thus, these hippocampal NMDARs are not essential for encoding or storing long-term, associative spatial memories. Instead, we demonstrate an important function of the hippocampus in using spatial knowledge to select between alternative responses that arise from competing or overlapping memories.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

1154  VOLUME 15 | NUMBER 8 | AUGUST 2012 nature neurOSCIenCe

a r t I C l e S

gliosis in 1-year-old mice (Fig. 1f and Supplementary Fig. 3). Thus, chronic removal of NMDARs in these hippocampal areas resulted in thinning of the stratum granulare in dentate gyrus, which might, in part, be related to a recently delineated function of NMDARs

in the adult dentate gyrus25. Grin1∆DGCA1 mice appeared normal and were visually indistinguishable from their littermate controls. However, they were hyperactive when placed in a novel environment (Supplementary Fig. 4).

Grin12lox

Grin1tm1RspTgCN12; TgLC1

TgCN12; TgLC1 TgCN12; TgLC1; Gt(ROSA)26Sor

TgCN12; TgLC1

TgCN12; TgLC1; Gt(ROSA)26Sor P75

Grin1�DGCA1

Grin1�DGCA1

Cre (–dox)

M123

Control

P28 P175 P175mfDG

Anti-Cre

Anti-Cre

CA1mf pi

P175

Grin1tm1Rsp∆

Grin

1 IS

HG

ria1

ISH

Grin1�DGCA1

100

80

60

40

20

0E0 P0

ControlNeuN Calbindin

DG

a b c

f

e

DG

Grin1�DGCA1Control

+ dox – doxP28

Cre

+ C

A1

cells

(%

)

OB CA1

DG2 mm

P45

Cre

Cre + NeuN

NeuN

d

DG

DG

CA1CA1

CA1

CA1

DG

P260–1.4 mm

–2.8 mm

–4.0 mm

OB

OB

OB

DG

97.2 ± 2.5

53.4 ± 19.1

19.2 ± 9.8 95 ± 2

98 ± 2

40 µm

CA1

CA1

Cre

Cre/NeuNNeuN

P150CA1

CA3

DG

Figure 1 Removal of NMDARs in dentate gyrus and CA1. (a) Genes to generate conditional GluN1 NMDAR knockout mice (Grin1∆DGCA1) by postnatal dox-controlled activation of TgCN12; TgLC1-encoded Cre in Grin1tm1Rsp mice (Grin12lox). Mice are treated with dox until P0. After dox withdrawal (– dox), Cre expression is activated. loxP sites, black triangles; exons, boxes; exons encoding membrane-inserted segments M1–M3, black. (b) Anti-Cre immunostaining of brain sections from dox-treated (until P0) TgCN12; TgLC1 mice at postnatal day (P) 28 and P175, and β-galactosidase staining (blue) of sections from TgCN12; TgLC1; Gt(ROSA)26Sor mice, age 8 months (mf, mossy fibers; pi, piriform cortex). Scale bars, 1 mm; bar in third panel from left applies to first and second. (c,d) Dox-regulated Cre expression in the brains of TgCN12; TgLC1 mice. (c) Expression of the nuclearly localized Cre in CA1 pyramidal cells, with dox removed from the drinking water of mothers at the day of delivery. At P28 (n = 2 mice), P45 (n = 2) and P150 (n = 6), we estimated the extent of Cre-expressing pyramidal neurons in the dorsal CA1 layer by immunostaining (right inset) with anti-Cre (green) counterstained with anti-NeuN (red); scale bar, 1 mm. E0, embryonic day 0. Left inset: Cre-activated β-galactosidase expression in CA1 and dentate gyrus (DG) of the hippocampus and the olfactory bulb (OB) in a brain section from TgCN12; TgLC1; Gt(ROSA)26Sor mice. (d) Left: horizontal brain sections of TgCN12; TgLC1 mice immunostained for Cre (green) and NeuN (red) at three different levels from bregma (−1.4, −2.8 and −4.0 mm), showing a yellow gradient of Cre plus NeuN immunoreactivity in the dorsal to ventral stratum pyramidale of CA1 but not in the DG. Scale bar, 1 mm. Right: high-magnification examples from sections on left of CA1 and DG neurons labeled with both anti-Cre and anti-NeuN (yellow). The percentage of Cre+ cells among >1,000 NeuN+ neurons was evaluated at various depths in dorsal and ventral hippocampus, in slices from four P260 mice, and is given as mean ± s.d. The transition between dorsal and ventral hippocampus in the rostro-caudal axis had the highest variance of Cre expressing NeuN+ cells (middle insets). Other parts of the hippocampal formation featured negligible numbers of Cre+ cells (<1% co-labeling). (e) In situ hybridization (ISH) with probes for Grin1 (encoding GluN1) and Gria1 (encoding GluA1). Scale bar, 1 mm. (f) NeuN and calbindin immunostainings. Scale bar, 1 mm. Mice in e,f were analyzed 3–5 weeks after they finished the last behavioral experiment.

Figure 2 Loss of functional NMDA receptors at CA3-to-CA1 synapses in the dorsal CA1 region. (a) Absence of NMDA responses in field recordings in acute brain slices from 8- to 12-month old Grin1∆DGCA1 mice. Blue traces, average field excitatory postsynaptic potentials (fEPSPs) elicited by three synaptic activations at 100 Hz in the presence of the AMPA receptor blocker 6,7-dinitroquinoxaline-2,3-dione (DNQX; 20 µM) in control and Grin1∆DGCA1 mice. Red traces are the corresponding responses following subsequent superfusion with dl-AP5 ((2R)-amino-5-phophonopentanoate) (50 µM) to block NMDA receptors. Mean representative traces (averaged across five stimulations at 0.1 Hz both before and after AP5) from a single slice from both a control and a Grin1∆DGCA1 mouse. Experiments were repeated in three control (n = 12 slices) and two Grin1∆DGCA1 mice (n = 8 slices). (b) Normalized (n-) and pooled fEPSP slopes evoked at CA1 stratum radiatum and stratum oriens synapses in slices from control and Grin1∆DGCA1 mice. Top inset: Schematic of two stimulation (stimul., red lines) and two recording electrodes (black lines) used for recordings in stratum oriens and stratum radiatum. Blue, hippocampal cell layers. Forty to 45 min after the last tetanization of the afferent fibers in the stratum radiatum in slices from five controls, the average slope of the fEPSP (open circles) was 1.60 ± 0.07 (mean ± s.e.m.; n = 18 slices) of the pre-tetanic value, whereas that in the untetanized stratum oriens pathway (triangles) was unchanged (0.97 ± 0.04). In slices from four Grin1∆DGCA1 mice, LTP was abolished (0.99 ± 0.03, n = 14 slices), and evoked responses were not significantly different from those in the untetanized pathway (1.03 ± 0.04; P = 0.35). (c) LTP at CA3–CA3 and CA3–CA1 synapses: normalized and pooled fEPSPs before and after LTP induction by a single electrode (top inset, stimul.) at CA3–CA3 and CA3–CA1 stratum radiatum synapses. In slices from three control mice, LTP was well developed 40–45 min after tetanization, with similar magnitudes in the two regions (CA3, 1.33 ± 0.12; CA1, 1.37 ± 0.11; n = 11 slices; P = 0.77). The same experiment performed on slices from three Grin1∆DGCA1 mice showed preserved LTP in CA3, whereas no LTP developed in CA1 (CA3, 1.20 ± 0.07; CA1, 0.94 ± 0.19; n = 8 slices; P = 0.01). All means ± s.e.m. Arrows, time points of tetanic stimulation. In b,c, bottom inset traces represent means of six consecutive synaptic responses in the tetanized pathway before and 45 min after tetanization.

a

b

cfEPSPs + DNQX

n-fE

PS

Ps

n-fE

PS

Ps

n-fE

PS

Ps

Stimul.

Stimul.

CA1

10 ms

10 ms

10 ms

10 ms

1 mV

2 mV

2 mV

2 mV

Control

Control

Control

0 20 40 min min0 20 40 60

Grin1�DGCA1

Grin1�DGCA1 Grin1�DGCA1

CA3–CA1

CA3–CA1

CA3–CA1CA3–CA3

CA1CA3

+ AP5– AP5

2

1

0

2

1

0

2

1

0

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nature neurOSCIenCe  VOLUME 15 | NUMBER 8 | AUGUST 2012 1155

a r t I C l e S

Electrophysiological analysisAs expected, no NMDAR-mediated responses could be evoked at CA3-to-CA1 synapses in dorsal hippocampal slices from adult Grin1∆DGCA1 mice (Fig. 2a). Extracellular field LTP experiments confirmed the loss of functional NMDARs in the Schaffer collateral/commissural pathway in dorsal CA1. In Grin1∆DGCA1 mice, no field LTP developed after four tetanizations (Fig. 2b).

In subsequent experiments, we placed a stimulation electrode in stratum radiatum, at the border between CA1 and CA3. Stimulation at 0.1 Hz elicited synaptic responses recorded simultaneously by two glass electrodes located in stratum radiatum in CA1 and CA3. In controls, LTP was well developed in both pathways 40–45 min after repeated tetanization, with similar magnitudes in the two subfields (Fig. 2c, top). In Grin1∆DGCA1 mice, LTP in the CA3 subfield was preserved, whereas CA3-to-CA1 LTP was absent (Fig. 2c, bottom), providing direct evidence for loss of NMDAR-mediated function in CA1. Thus, Grin1∆DGCA1 mice provide a means of assessing the hippocampal NMDAR-dependent synaptic plasticity and memory hypothesis. We therefore assessed associative, long-term spatial memory in Grin1∆DGCA1 mice using two classic, hippocampus-dependent tests (Supplementary Fig. 5).

Spatial reference memory in the Morris water mazeMice with hippocampal lesions were unable to acquire memory of the platform location in the fixed-location, hidden-platform version of the water maze (Supplementary Fig. 5a–c). In contrast, experi-mentally naive Grin1∆DGCA1 mice learned the spatial location of the platform as well as controls did (Fig. 3). Mice exhibited reduced path lengths to find the platform as training proceeded, with no impairment in Grin1∆DGCA1 mice (ANOVA; main effect of block, F8,176 = 62.68, P < 0.0001; no main effect of genotype, F < 1; no geno-type × block interaction, F8,176 = 1.64, P > 0.10; Fig. 3a). During probe tests in which the platform was removed from the pool, hippo-campally lesioned mice characteristically searched equally across all four quadrants of the pool (Supplementary Fig. 5c), whereas in the first probe test (24 h after trial 24) both controls and Grin1∆DGCA1 mice

exhibited strong spatial memory and searched selectively at the train-ing platform location (Fig. 3b). ANOVA revealed only a significant main effect of quadrant (F2,66 = 49.66; P < 0.05). In the second probe test (24 h after trial 36), Grin1∆DGCA1 mice showed stronger spatial memory and spent significantly more time in the training quadrant than controls (Fig. 3c). ANOVA revealed a main effect of quad-rant (F2,66 = 68.74; P < 0.05) and a genotype × quadrant interaction (F2,66 = 3.27; P < 0.05). There was a significant group difference in time spent in the training quadrant (t-test; t22 = 2.24; P < 0.05).

On the next day the platform was moved to the opposite quadrant of the pool for a further 3 d of training. Whereas initial acquisition had been normal in Grin1∆DGCA1 mice, during reversal learning they were significantly impaired, taking longer paths to the new platform position (ANOVA; main effect of genotype, F1,22 = 13.41, P < 0.005; main effect of block, F2,44 = 36.54, P < 0.0001; no genotype × block interaction, F2,44 = 2.24, P > 0.10; Fig. 3a).

Spatial reference memory on the radial mazeAlthough Grin1∆DGCA1 mice displayed normal acquisition in the water maze, they were, like mice with hippocampal lesions (Supplementary Fig. 5d), substantially impaired in learning a radial maze task, in which they had to discriminate between three always-baited arms and three never-baited arms (Fig. 4). Grin1∆DGCA1 mice made more reference memory errors (ANOVA; main effect of genotype, F1,21 = 30.42, P < 0.0001; main effect of block, F11,231 = 16.08, P < 0.0001; no genotype × block interaction, F11,231 = 1.03, P > 0.40; Fig. 4a). We also compared the number of erroneous entries into the isolated (single) non-rewarded arm and the pair of adjacent non-rewarded arms (allowing for the number of arms) (Fig. 4b). Mice were more likely to erroneously enter the isolated (single) non-rewarded arm than one of the pair of non-rewarded arms (ANOVA; main effect of error type,

Reversal

Grin1∆DGCA1

Control12

Pat

h le

ngth

(m

)

Block (4 trials)

a Acquisition

TT1 TT28

12

4Pat

h le

ngth

(m

)

00 1 2 3 4 5

Block (4 trials)6 7 8 9 1 2 3

8

4

0

b80 6

Transfer test 1 (TT1), after 24 trials

6040

Tim

e (%

)

20

Contro

l

Contro

l0

4

Ann

ulus

cro

ssin

gs

2

0 GG G

Grin1∆DGCA1

Grin1∆DGCA1

c80

*6

Transfer test 2 (TT2), after 36 trials

Contro

lG G GG

6040

Tim

e (%

)

200

4

2

Ann

ulus

cro

ssin

gs

0

Grin1∆DGCA1

Contro

l

Grin1∆DGCA1

G

Figure 3 Associative long-term spatial reference memory in the Morris water maze. (a) Similar path lengths for controls (n = 12) and Grin1∆DGCA1 mice (n = 12) during acquisition of memory in the fixed-location, hidden-escape-platform water maze task (four trials per block). When the platform was moved to the opposite quadrant of the pool (reversal), the path lengths of Grin1∆DGCA1 mice were significantly longer than those of controls (P < 0.005). All means ± s.e.m. (b,c) Long-term memory performance during probe trials (transfer tests) after 6 and 9 training blocks. Percentage time spent and numbers of annulus (would-be platform) crossings in each quadrant (left to right: adjacent left, goal (G), adjacent right, opposite) for controls and Grin1∆DGCA1 mice. In transfer test 2 (TT2), Grin1∆DGCA1 mice searched longer in the training quadrant than controls (*P < 0.05). Dashed line, chance performance. All data are means ± s.e.m.

3.0a b

40

30

20

10

Acquisition

AdjAdj

+

––

–

+

+

Sin

ControlGrin1�DGCA1

ControlGrin1�DGCA1

2.5

2.0

1.5

Err

ors

Error types

Err

ors

1.0

0.5

00 1 2 3 4 5 6

Block (4 trials)

7 8 9 10 11 12Adj/

2 SinAdj/

2 Sin

Figure 4 Associative long-term spatial reference memory on the radial maze task. (a) Inset: Mice were trained to discriminate between three rewarded (+) and three non-rewarded (−) arms, two of them adjacent (Adj) and one single (Sin). Errors per trial for controls (n = 12) and Grin1∆DGCA1 mice (n = 11) across 12 blocks of testing (4 trials per block). Grin1∆DGCA1 mice were impaired at discriminating between the rewarded and non-rewarded arms, making significantly more spatial reference memory errors (P < 0.0001). (b) Error types during spatial reference memory acquisition. Total number of errors into the single non-rewarded arm and the total number of errors into the pair of adjacent non-rewarded arms. The number of errors into the pair of adjacent non-rewarded arms was divided by two to correct for the number of arms. All means ± s.e.m.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

1156  VOLUME 15 | NUMBER 8 | AUGUST 2012 nature neurOSCIenCe

a r t I C l e S

F1,21 = 7.06, P < 0.02). However, this pattern was the same for both genotypes (genotype × error type interaction, F < 1, P > 0.90).

The dissociation between these two classic spatial reference memory tests (water maze and radial maze) did not reflect differences in senso-rimotor, motivational or arousal demands of these tasks26. The behav-ioral difference between genotypes was also present in both appetitive and aversive (swim escape) versions of a three-arm radial maze (Y-maze) reference memory task. In both versions of this Y-maze task, Grin1∆DGCA1 mice were impaired during acquisition (Supplementary Fig. 6). Furthermore, the two groups learned an appetitively moti-vated visual discrimination task (gray versus black-and-white-striped goal arms) at similar rates (Fig. 5). ANOVA revealed a main effect of block (F8,176 = 36.18; P < 0.0001), but no main effect of genotype (F < 1) and no genotype × block interaction (F8,176 = 1.07; P > 0.30). The groups were also indistinguishable during subsequent reversal of this non-spatial task (Supplementary Fig. 7).

Spatial discrimination using a beacon water maze taskAs differences in sensorimotor, motivational or arousal demands can-not account for the dissociation between water maze and radial maze performance in Grin1∆DGCA1 mice, the dissociation is likely instead to reflect the different psychological demands of these tasks. On the radial maze, all arms have the same physical appearance (the intra-maze cues are constant). They have the same transparent plastic door, the same gray floor and sidewalls and the same food well. When mice find food at the ends of the arms, these common features become associated with reward, so mice will tend to run down arms expecting food. However, these common intra-maze cues are present in both

rewarded and non-rewarded arms and are therefore ambiguous. To inhibit the tendency to run down non-rewarded arms, and so discrim-inate successfully between baited and never-baited arms, the mice must therefore use the arm-specific, extra-maze spatial cues to select the correct response (run versus don’t run) for each arm. We hypo-thesized that it is this ability to use spatial information to disambiguate competing or overlapping memories (a form of pattern separation), and thereby inhibit inappropriate associative, conditioned behaviors, that is impaired in Grin1∆DGCA1 mice on this task.

To test this possibility, we trained experimentally naive mice on a water maze task in which they were required to discriminate between two visibly identical beacons (black spheres), depending on their spa-tial locations (Fig. 6a)27,28. One beacon was located over the platform. The other, ‘decoy’ beacon was at the point-symmetrically opposite location in the pool and provided no escape. Both beacons remained in fixed spatial locations. Mice were placed in the pool, either close to the correct beacon (S+ position), close to the decoy beacon (S− posi-tion), or equidistant between the two beacons (Fig. 6b).

Grin1∆DGCA1 mice were impaired in their spatial choices between the two beacons, despite knowing the platform’s spatial location. During pre-training with a single beacon in a variable spatial loca-tion, all mice rapidly learned to swim toward the beacon and escape onto the platform. During subsequent spatial discrimination training, both groups then learned to select the correct beacon according to its spatial location as shown by the gradual improvement in first-choice accuracy (ANOVA; main effect of block, F4,84 = 24.52, P < 0.0001; no main effect of genotype, F1,21 = 2.88, P > 0.10; no genotype × block interaction, F < 1, P > 0.70; Fig. 6c). Furthermore, when probe tests were performed, with both beacons and the platform removed from the pool, both groups equally preferred the quadrant where the plat-form had been, demonstrating spatial learning (Fig. 6d; probe test 1 ANOVA: effect of quadrant, F2,63 = 19.29, P < 0.05; no genotype × quadrant interaction, F < 1, P > 0.80; time spent in the training quadrant, t-test, t < 1, P > 0.80; probe test 2 ANOVA: main effect of quadrant, F2,63 = 19.90, P < 0.05; no genotype × quadrant interaction,

Beacons

First-choice accuracy Transfer testsTT1 (after 72 trials)

TT2 (after 120 trials)

Control

G

G

TT1 TT290

90705030

40

20

0

40

20

0

80

70

60

Cor

rect

(%

)C

orre

ct (

%)

Tim

e (%

)T

ime

(%)50

400 1 2 3

Block (24 trials)4 5

Control

Grin1�DGCA1

Control

*

S+

S+

Equid.

Equid.

Block Block Block1 2 3 4 5 1 2 3 4 5 1 2 3 4 5S–

S–Grin1�DGCA1

Grin1�DGCA1

Equid.

Equid.

S–

S–

S+

S+

Side viewa b

dc

e

Top view

G

G

Figure 6 Spatial memory and choice performance on the spatial-discrimination beacon water maze task. (a) Schema and side view of the beacon water maze task. (b) Top view indicating the six possible start sites and the positions of the beacons (dotted line, platform; shaded circle, no platform). Equid., start site equidistant between the beacons. (c) Both controls (n = 11) and Grin1∆DGCA1 mice (n = 12) learned to choose the correct beacon in terms of first-choice accuracy (24 trials per block). (d) On probe tests (TT1 and TT2; beacons and platform removed), both groups showed a strong preference for the goal quadrant (from left to right: adjacent left quadrant, goal (G), adjacent right, opposite). (e) Left: Grin1∆DGCA1 mice were impaired at choosing the correct beacon when starting from a point close to the incorrect beacon (S−; *P < 0.025), but not when starting from the S+ or equidistant start locations. Right: percentage correct choices across five blocks of training (eight trials per block) on trials starting from the S+, the equidistant and the S− position. Dashed lines, chance performance. All means ± s.e.m.

100

90

80

70

Cor

rect

(%

)

60

50

ControlGrin1�DGCA1

40

300 1 2 3 4 5 6 7 8 9

Block

+ – – +

SS

Figure 5 Preserved visual discrimination learning in Grin1∆DGCA1 mice. Left: in the visual-discrimination task, both controls (n = 13) and Grin1∆DGCA1 mice (n = 11) learned to associate floor and wall color (gray versus black-and-white stripes) with a reward independent of its spatial position in the arms of the maze. Right: T-maze with the start arm (S) and rewarded (+) and non-rewarded (−) goal arms, which could be arranged in either of two possible configurations. There were 9 blocks of training (10 trials per block). Mean ± s.e.m. percentage correct choices. Dashed line, chance performance.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nature neurOSCIenCe  VOLUME 15 | NUMBER 8 | AUGUST 2012 1157

a r t I C l e S

F2,63 = 2.07, P > 0.10; time spent in the training quadrant, t-test, t21 = 1.75, P = 0.09). Thus, Grin1∆DGCA1 mice learned the location of the platform as well as the controls, successfully replicating the finding of preserved associative spatial memory from our first experiment.

However, when we examined first-choice accuracy in terms of starting position, Grin1∆DGCA1 mice were impaired at choosing the correct beacon when started from close to the decoy beacon (S−), but not when starting from either of the other start locations (S+ or equidistant; Fig. 6e). Not surprisingly, given that they had learned to swim toward the beacon as a means of escape during pre-training, all mice initially approached the first beacon that they encountered, which resulted in sub-chance performance on S− trials and above-chance performance on S+ trials. Mice then gradu-ally learned to inhibit this approach response depending on the spatial location of the beacon. ANOVA of first-choice accuracy also revealed a main effect of start position, which was due to poorer performance when mice started close to the S− beacon and better performance when they started close to the S+ position (F2,42 = 233.50; P < 0.0001). There was also a significant genotype × start position interaction (F2,42 = 4.24; P < 0.025). Further investigation, using analysis of simple main effects, confirmed that Grin1∆DGCA1 mice were impaired when trials started from close to the S− beacon (F1,21 = 5.92; P < 0.025) but not when trials started close to the S+ beacon (F < 1; P > 0.80), nor from equidistant between the beacons (F1,21 = 1.86; P > 0.10). Separate ANOVAs revealed significant improve-ments with training for each start position (main effect of block; S− trials, F4,84 = 8.24, P < 0.0001; equidistant trials, F4,84 = 24.50, P < 0.0001; S+ tri-als, F4,84 = 5.63, P < 0.0005). By the end of training, the control mice were exhibiting above chance levels of performance from all start positions (for example, on S− trials only, one-group t-test, P < 0.05).

Because mice sometimes swam under the S− beacon and then re-emerged before again swimming under the S− beacon, we also analyzed total errors, with each repeated choice adding to a cumu-lated score for the trial. This revealed a main effect of genotype (F1,21 = 18.53; P < 0.0005) and an interaction between genotype and starting position (F2,42 = 5.18; P < 0.01; Supplementary Fig. 8). Subsequent analysis of the genotype × start position interaction using simple main effects confirmed that Grin1∆DGCA1 mice were impaired on S− trials (F1,21 = 17.02; P < 0.001) and unimpaired on S+ trials (F < 1; P > 0.30). On the total errors measure, they were now signifi-cantly impaired on equidistant trials as well (F1,21 = 5.39; P < 0.05), presumably because, if they did approach the wrong beacon, they were then more likely than controls to persist in that response.

Non-spatial discrimination using a beacon water maze taskGrin1∆DGCA1 mice performed as well as controls on an analogous, non-spatial version of the task in which they had to choose between two visually distinctive beacons27 (gray funnel versus black-and-white striped cylinder; Fig. 7a,b). Mice learned the visual discrimination task, displaying increased first-choice accuracy for the S+ beacon and

reduced total errors as training proceeded, but there were no group differences (Fig. 7c,d and Supplementary Fig. 9). As in the spatial discrimination task, mice found the task harder (they were less accu-rate with their first choices and made more errors) when starting from close to the S− beacon. Conversely, they performed more efficiently when trials started from close to the S+ beacon. ANOVA of first-choice accuracy revealed a main effect of block, demonstrating learn-ing of the task (F5,75 = 27.80; P < 0.0001; Fig. 7c). There was also a main effect of start position, which was due to mice being more likely to choose incorrectly when starting from close to the S− beacon and more likely to choose correctly when starting from closer to the S+ beacon (F2,30 = 48.20; P < 0.0001; Fig. 7c right and Fig. 7d). Notably, however, there was no main effect of genotype or significant interaction involving genotype (ANOVA; genotype, F < 1, P > 0.60; genotype × block and genotype × start position, both F < 1, P > 0.60; genotype × block × start position, F10,150 = 1.09, P > 0.30). Similar results were obtained when analyzing total errors (Supplementary Fig. 9). Finally, there was no impairment in Grin1∆DGCA1 mice during subsequent reversal of the non-spatial beacon task (Supplementary Fig. 10).

DISCUSSIONThe present results show that NMDARs on dentate gyrus and dorsal CA1 principal cells are not essential neuronal underpinnings of hippocampus-dependent, associative, spatial reference memory acquisition or storage. Instead they may be critical for using spatial information to guide selection between alternative responses. Despite ablation of NMDARs from both dentate gyrus granule cells and dor-sal CA1 pyramidal cells, and also some dentate gyrus granule cell loss, Grin1∆DGCA1 mice still learned the hippocampus-dependent, spatial ref-erence memory version of the water maze as well as controls. This was despite the loss of NMDAR currents and LTP in dorsal CA1 pyramidal cells in these animals. Although our manipulation was not confined to dorsal CA1 but also affected ventral CA1 and the entire dentate gyrus, these additional consequences cannot explain the lack of a deficit on the standard water maze task. Furthermore, although there was a gradient in ventral hippocampus of residual CA1 pyramidal cells that retained

Figure 7 Normal performance of Grin1∆DGCA1 mice on the visual-discrimination beacon water maze task. (a) Schema and side view of the non-spatial beacon water maze task. (b) Top view of the six possible start positions. (c) Both controls (n = 8) and Grin1∆DGCA1 mice (n = 9) learned to choose the correct beacon. First-choice accuracy improved with training for both groups across six blocks of training (24 trials per block; c, left). First-choice accuracy across all training trials from different starting positions showed that all mice were less accurate at choosing the correct beacon when starting from S− (c, right). (d) Performance across six blocks of training on trials starting from S+, equidistant and S− positions. All mice performed less accurately when starting from S−. There was no difference between controls and Grin1∆DGCA1 mice across six blocks of training for any of the start positions (eight trials per block). Dashed lines, chance performance. All means ± s.e.m.

Side view

Beacons

First-choice accuracy90

First-choice accuracy

Top view

Equid.

Equid.

Equid.

S+

S+

S+

S–

S–

S–

100

80

60

40

20

80

70

60

50

40

90

0 1 2 3 4Block

Block Block Block

S+ start position S– start positionEquid. start postion

Cor

rect

(%

)

Cor

rect

(%

)

Cor

rect

(%

)

5 6

ControlGrin1�DGCA1

Control

Grin1�DGCA1

70

50

30

0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6

a

c

d

b

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

1158  VOLUME 15 | NUMBER 8 | AUGUST 2012 nature neurOSCIenCe

a r t I C l e S

NMDARs, it is unlikely that these could support water maze perform-ance because lesion data from both rats and mice show that this ventral region alone is insufficient to support spatial learning on this task18–20.

In contrast, Grin1∆DGCA1 mice were unable to acquire the hippo-campus-dependent, spatial reference memory radial maze task. This dissociation between the two classic tests of associative long-term spatial memory is not due to different sensorimotor or motivational task demands but instead reflects the different psychological proc-esses involved. To investigate this dissociation further, we ran a variant of the water maze task in which mice had to discriminate spatially between two visually identical beacons. Grin1∆DGCA1 mice were less able than controls to withhold a response to the incorrect, decoy beacon despite an equal knowledge of the spatial location of the platform. This did not reflect a generalized increase in approach tendencies. When we used two visually distinct, and hence unambigu-ous, beacons, Grin1∆DGCA1 mice were unimpaired. Thus, an account based on how likely mice were to swim to any beacon, or based on task difficulty, cannot explain these data. Instead the impairment in the spatial task arose specifically as a result of the ambiguity evoked by two identical beacons. Although previous studies have revealed dissociations within the domain of associative, long-term spatial memory29, here we show that even when spatial navigation is normal in Grin1∆DGCA1 mice, they are unable to use that spatial knowledge to select appropriately between competing response options.

The behavioral consequences in Grin1∆DGCA1 mice were distinct from those seen after NMDAR deletion restricted to just dentate gyrus gran-ule cells14. Grin1∆DG mice show normal long-term associative spatial reference memory on the radial maze (although they do exhibit spatial working memory impairments). The Grin1∆DGCA1 mice in the present study were impaired on the associative, long-term reference memory radial maze task, and one might therefore conclude that it is the gene ablation in CA1 pyramidal cells that is specifically responsible, although we cannot rule out a cumulative phenotype in dentate gyrus and CA1. Nevertheless, the very selective behavioral impairments exhibited by Grin1∆DGCA1 mice, combined with the preserved ability to learn the spatial location of the platform in the water maze, provide key informa-tion about the algorithm being performed by the hippocampus.

Our data dispute previous claims that NMDARs in dorsal CA1 are essential for associative spatial memory formation in the water maze7, based on conditional Grin1 knockout by means of the Tg29-1 transgenic mouse line, which was thought to be confined to dorsal CA1 pyramidal cells. Potential differences between the two studies include genetic background, temporal onset of the genetic deletion, the dorsal–ventral extent of GluN1 ablation in the hippocampus and the age of the mice. The most likely explanation, however, is the lack of hippocampal selectivity of the Grin1 knockout, which more recent studies of the Tg29-1 line have demonstrated spreads into cortical areas9–13. Subsequent publications report a clear reduction in cortical GluN1 expression, measured by in situ hybridization and immuno-cytochemistry, as early as 2 months of age if not sooner11,13. Indeed, the Tg29-1 line expresses Cre in cortex as early as 6 weeks after birth9. Notably, the Tg29-1 mice with loxP-flanked Grin1 alleles were also impaired on a visually cued version of the water maze task7, demon-strating a hippocampus-independent learning impairment and indi-cating an effect of Grin1 gene inactivation outside the hippocampus.

Combined with our current data, these results suggest that NMDARs, either elsewhere in the extended hippocampal formation, such as entorhinal cortex30 or subiculum31, or across the wider cortical mantle, could underlie associative spatial memory performance in the water maze. Notably, a similar conclusion arises from studies with conditional Grin2b knockout mice. Whereas ablation of the GluN2B subunit in

both hippocampus and cortex impairs water maze learning12, deletion of the subunit specifically from dorsal CA1 pyramidal cells and dentate gyrus granule cells has no effect on this task22. Thus, cortical NMDARs are likely to be important in spatial memory acquisition.

At first glance, the current results also seem at odds with pharmaco-logical studies in rats. Intracerebroventricular infusion of the NMDAR antagonist dl-AP5 impairs learning in the reference memory version of the water maze4. Of course, intracerebroventricular infusion will affect both hippocampal and extra-hippocampal NMDARs. Even so, subse-quent experiments found that water maze acquisition is not impaired by NMDAR blockade using dl-AP5 if rats have first been trained before drug treatment on a spatial task in a different room5,6. This is consistent with the present data in demonstrating that hippocampal NMDARs are not necessary (i) for spatial navigation per se, nor (ii) for forming a long-term association between a particular spatial location and the platform. In a further parallel with Grin1∆DGCA1 mice in the present study, despite spatial pre-training, dl-AP5–treated rats are impaired during spatial reversal in the water maze when the location of the platform is switched to a new place in the same, familiar environment32.

Thus, the role of hippocampal NMDARs lies not in associative mem-ory formation but elsewhere. Although NMDAR deletion in dentate gyrus and dorsal CA1 principal cells does not prevent encoding of long-term spatial memories, it does affect the use of spatial informa-tion to disambiguate overlapping or competing associative memories. It has been suggested that the hippocampus is a key component of a comparator system to detect mismatch or conflict33,34. Human func-tional magnetic resonance imaging experiments35–37 and both electro-physiological unit recording38–40 and lesion studies41 in rodents have implicated the hippocampus, and particularly the CA1 subfield, in the detection and/or resolution of associative mismatch and conflict, such as might occur between an expectation based on information retrieved from long-term memory and the current state of the perceptual world. This comparator function is not, however, a reward prediction error signal that determines the extent of associative learning42. Instead, one key output of this hippocampal mismatch detection system is behavio-ral inhibition of ongoing activity33,34, modifying within-trial perform-ance without affecting the progressive changes in associative learning from trial to trial. A failure in this process likely underlies the deficit in Grin1∆DGCA1 mice when the platform was moved to the opposite quadrant of the pool during the first water maze experiment (Fig. 3a; see also ref. 32). Notably, Grin1∆DGCA1 mice were not impaired in either of two non-spatial reversal paradigms. This excludes an explanation based on a general tendency to perseverate all learned responses.

The present results also demonstrate a new but related role for hippocampal NMDARs in resolving interference between competing long-term memories, so enabling behavioral inhibition of inappropri-ately cued or conditioned responses on the basis of spatial or contex-tual information. This key deficit in Grin1∆DGCA1 mice manifested as an inability to withhold a response to the decoy beacon, despite know-ing the spatial location of the platform. Critically, this did not reflect a general inability to inhibit all approach responses, as evidenced by the normal performance of Grin1∆DGCA1 mice on the visual discrimi-nation beacon task, irrespective of start position. Instead, the deficit occurred when a conflict arose as the result of ambiguity between competing long-term memories associated with two visibly identical beacons. This same account can explain the radial maze deficit in Grin1∆DGCA1 mice, which results from the ambiguity that arises from the identical physical appearance of all six arms of the maze.

It has repeatedly been suggested that an important function of the hip-pocampus is pattern separation to disambiguate overlapping inputs43–45. On this occasion the ambiguity arises, not from overlapping extra-maze

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nature neurOSCIenCe  VOLUME 15 | NUMBER 8 | AUGUST 2012 1159

a r t I C l e S

spatial cues, but from the similarity of the maze arms or the visibly identi-cal beacons and from the association of these cues with conflicting out-comes. Thus, at a psychological level, NMDARs in the hippocampus may serve to resolve conflict or ambiguity as a result of detecting mismatch or uncertainty rather than subserving learning by detecting coincidence.

METHODSMethods and any associated references are available in the online version of the paper.

Accession Codes. MGI: Grin1tm1Rsp and Gt(ROSA)26Sor.

Note: Supplementary information is available in the online version of the paper.

ACknowledgmentSWe thank U. Amtmann, H. Monyer and I. Bertocchi for in situ hybridization, and W. Tang and D. Arcos Diaz for recombinant adeno-associated virus injections. We thank R. Deacon for assistance with lesion surgeries. This work was supported by a grant from the European Commission to the EUSynapse project (LSHM-CT-2005-019055) to P.H.S., Wellcome Trust Senior Research Fellowships 074385 and 087736 to D.M.B., a grant from the German Research Foundation (SFB636/A4; BMFT:NGFN/SP10, DFG-GA 427/8-1) to R.S. and a grant from the Bauer-Stiftung to T.B. Ø.H. and V.J. were supported by the Letten Foundation.

AUtHoR ContRIBUtIonSR.S. and P.H.S. delineated the genetic concept. T.B., I.S. and R.S. generated and molecularly analyzed the Grin1∆DGCA1 mice. Ø.H. and V.J. performed the electrophysiological analysis. D.M.B. designed the behavioral experiments. D.M.B., D.J.S. and A.T. performed behavioral analyses. D.M.B., T.B., R.S., J.N.P.R. and P.H.S. wrote the manuscript.

ComPetIng FInAnCIAl InteReStSThe authors declare no competing financial interests.

Published online at http://www.nature.com/doifinder/10.1038/nn.3166. Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Bliss, T.V. & Collingridge, G.L. A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361, 31–39 (1993).

2. Martin, S.J., Grimwood, P.D. & Morris, R.G. Synaptic plasticity and memory: an evaluation of the hypothesis. Annu. Rev. Neurosci. 23, 649–711 (2000).

3. Neves, G., Cooke, S.F. & Bliss, T.V. Synaptic plasticity, memory and the hippocampus: a neural network approach to causality. Nat. Rev. Neurosci. 9, 65–75 (2008).

4. Morris, R.G., Anderson, E., Lynch, G.S. & Baudry, M. Selective impairment of learning and blockade of long-term potentiation by an N-methyl-d-aspartate receptor antagonist, AP5. Nature 319, 774–776 (1986).

5. Bannerman, D.M., Good, M.A., Butcher, S.P., Ramsay, M. & Morris, R.G. Distinct components of spatial learning revealed by prior training and NMDA receptor blockade. Nature 378, 182–186 (1995).

6. Saucier, D. & Cain, D.P. Spatial learning without NMDA receptor-dependent long-term potentiation. Nature 378, 186–189 (1995).

7. Tsien, J.Z., Huerta, P.T. & Tonegawa, S. The essential role of hippocampal CA1 NMDA receptor-dependent synaptic plasticity in spatial memory. Cell 87, 1327–1338 (1996).

8. Tsien, J.Z. et al. Subregion- and cell type-restricted gene knockout in mouse brain. Cell 87, 1317–1326 (1996).

9. Wiltgen, B.J. et al. A role for calcium-permeable AMPA receptors in synaptic plasticity and learning. PLoS ONE 5, e12818 (2010).

10. Hoeffer, C.A. et al. Removal of FKBP12 enhances mTOR-Raptor interactions, LTP, memory, and perseverative/repetitive behavior. Neuron 60, 832–845 (2008).

11. Fukaya, M., Kato, A., Lovett, C., Tonegawa, S. & Watanabe, M. Retention of NMDA receptor NR2 subunits in the lumen of endoplasmic reticulum in targeted NR1 knockout mice. Proc. Natl. Acad. Sci. USA 100, 4855–4860 (2003).

12. Brigman, J.L. et al. Loss of GluN2B-containing NMDA receptors in CA1 hippocampus and cortex impairs long-term depression, reduces dendritic spine density, and disrupts learning. J. Neurosci. 30, 4590–4600 (2010).

13. Rondi-Reig, L. et al. Impaired sequential egocentric and allocentric memories in forebrain-specific-NMDA receptor knock-out mice during a new task dissociating strategies of navigation. J. Neurosci. 26, 4071–4081 (2006).

14. Niewoehner, B. et al. Impaired spatial working memory but spared spatial reference memory following functional loss of NMDA receptors in the dentate gyrus. Eur. J. Neurosci. 25, 837–846 (2007).

15. McHugh, T.J. et al. Dentate gyrus NMDA receptors mediate rapid pattern separation in the hippocampal network. Science 317, 94–99 (2007).

16. Murray, A.J. et al. Parvalbumin-positive CA1 interneurons are required for spatial working but not for reference memory. Nat. Neurosci. 14, 297–299 (2011).

17. Dillon, G.M., Qu, X., Marcus, J.N. & Dodart, J.C. Excitotoxic lesions restricted to the dorsal CA1 field of the hippocampus impair spatial memory and extinction learning in C57BL/6 mice. Neurobiol. Learn. Mem. 90, 426–433 (2008).

18. Logue, S.F., Paylor, R. & Wehner, J.M. Hippocampal lesions cause learning deficits in inbred mice in the Morris water maze and conditioned-fear task. Behav. Neurosci. 111, 104–113 (1997).

19. Bannerman, D.M. et al. Double dissociation of function within the hippocampus: a comparison of dorsal, ventral, and complete hippocampal cytotoxic lesions. Behav. Neurosci. 113, 1170–1188 (1999).

20. Moser, M.B., Moser, E.I., Forrest, E., Andersen, P. & Morris, R.G. Spatial learning with a minislab in the dorsal hippocampus. Proc. Natl. Acad. Sci. USA 92, 9697–9701 (1995).

21. Pothuizen, H.H., Zhang, W.N., Jongen-Relo, A.L., Feldon, J. & Yee, B.K. Dissociation of function between the dorsal and the ventral hippocampus in spatial learning abilities of the rat: a within-subject, within-task comparison of reference and working spatial memory. Eur. J. Neurosci. 19, 705–712 (2004).

22. von Engelhardt, J. et al. Contribution of hippocampal and extra-hippocampal NR2B-containing NMDA receptors to performance on spatial learning tasks. Neuron 60, 846–860 (2008).

23. Shimshek, D.R. et al. Enhanced odor discrimination and impaired olfactory memory by spatially controlled switch of AMPA receptors. PLoS Biol. 3, e354 (2005).

24. Suchanek, B., Seeburg, P.H. & Sprengel, R. Tissue specific control regions of the N-methyl-D-aspartate receptor subunit NR2C promoter. Biol. Chem. 378, 929–934 (1997).

25. Tashiro, A., Sandler, V.M., Toni, N., Zhao, C. & Gage, F.H. NMDA-receptor-mediated, cell-specific integration of new neurons in adult dentate gyrus. Nature 442, 929–933 (2006).

26. Lyon, L. et al. Fractionation of spatial memory in GRM2/3 (mGlu2/mGlu3) double knockout mice reveals a role for group II metabotropic glutamate receptors at the interface between arousal and cognition. Neuropsychopharmacology 36, 2616–2628 (2011).

27. Morris, R.G., Hagan, J.J. & Rawlins, J.N. Allocentric spatial learning by hippocampectomised rats: a further test of the “spatial mapping” and “working memory” theories of hippocampal function. Q. J. Exp. Psychol. B 38, 365–395 (1986).

28. Uekita, T. & Okaichi, H. Pretraining does not ameliorate spatial learning deficits induced by intrahippocampal infusion of AP5. Behav. Neurosci. 123, 520–526 (2009).

29. Brun, V.H. et al. Place cells and place recognition maintained by direct entorhinal-hippocampal circuitry. Science 296, 2243–2246 (2002).

30. Steffenach, H.A., Witter, M., Moser, M.B. & Moser, E.I. Spatial memory in the rat requires the dorsolateral band of the entorhinal cortex. Neuron 45, 301–313 (2005).

31. Morris, R.G., Schenk, F., Tweedie, F. & Jarrard, L.E. Ibotenate lesions of hippocampus and/or subiculum: dissociating components of allocentric spatial learning. Eur. J. Neurosci. 2, 1016–1028 (1990).

32. Morris, R.G., Davis, S. & Butcher, S.P. Hippocampal synaptic plasticity and NMDA receptors: a role in information storage? Phil. Trans. R. Soc. Lond. B 329, 187–204 (1990).

33. Vinogradova, O.S. Functional organization of the limbic system in the process of registration of information: facts and hypotheses. in The Hippocampus vol. 2 (eds. Isaacson, R.I. & Pribram, K.H.) 3–69 (Plenum, New York, 1975).

34. Gray, J.A. & McNaughton, N. The Neuropsychology of Anxiety (Oxford Univ. Press, Oxford, 2000).

35. Ploghaus, A. et al. Learning about pain: the neural substrate of the prediction error for aversive events. Proc. Natl. Acad. Sci. USA 97, 9281–9286 (2000).

36. Kumaran, D. & Maguire, E.A. Match mismatch processes underlie human hippocampal responses to associative novelty. J. Neurosci. 27, 8517–8524 (2007).

37. Kumaran, D. & Maguire, E.A. An unexpected sequence of events: mismatch detection in the human hippocampus. PLoS Biol. 4, e424 (2006).

38. O’Keefe, J. & Nadel, L. The Hippocampus as a Cognitive Map (Clarendon, Oxford, 1978).

39. O’Keefe, J. Place units in the hippocampus of the freely moving rat. Exp. Neurol. 51, 78–109 (1976).

40. Fyhn, M., Molden, S., Hollup, S., Moser, M.B. & Moser, E. Hippocampal neurons responding to first-time dislocation of a target object. Neuron 35, 555–566 (2002).

41. Honey, R.C., Watt, A. & Good, M. Hippocampal lesions disrupt an associative mismatch process. J. Neurosci. 18, 2226–2230 (1998).

42. Hollerman, J.R. & Schultz, W. Dopamine neurons report an error in the temporal prediction of reward during learning. Nat. Neurosci. 1, 304–309 (1998).

43. Marr, D. Simple memory: a theory for archicortex. Phil. Trans. R. Soc. Lond. B 262, 23–81 (1971).

44. Rolls, E.T. & Treves, A. Neural Networks and Brain Function (Oxford Univ. Press, Oxford, 1998).

45. Sahay, A., Wilson, D.A. & Hen, R. Pattern separation: a common function for new neurons in hippocampus and olfactory bulb. Neuron 70, 582–588 (2011).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nature neurOSCIenCe doi:10.1038/nn.3166

ONLINE METHODSAnimal experimentation. Molecular, biochemical experiments and genetic manipulations of mouse embryos were licensed by the regional council in Karlsruhe, Germany, under project licenses 37-9185.81/G71-19 and G74-07. Electrophysiological studies were conducted according to the Norwegian Animal Welfare Act and European Union’s Directive 86/609/EEC. Behavioral experiments were in accordance with the UK Animals, Scientific Procedures Act (1986), under project licenses PPL 30/1989 and PPL 30/2561. Efforts were made to minimize numbers of animals used. Data from about 80 male and female mice are presented in this study.

generation of Grin1∆DGCA1 mice. Gene-targeted Grin12lox mice with loxP5171 (ref. 46) flanking Grin1 exons 11–18 (Grin1tm1Rsp) were generated as described previously14. Grin1∆DGCA1 mice were homozygous for the floxed Grin1tm1Rsp allele and carried the two transgenes TgCN12 and TgLC1. TgCN12 expresses the tetracycline-dependent transcriptional transactivator, itTA, from a NRSE element–modified Camk2a promoter, and TgLC1 encodes the itTA-controlled Cre recombinase and luciferase24. Controls were Grin12lox littermates lacking both or either one of the transgenes of lines TgCN12 or TgLC1. Dox (50 mg/l in drinking water) was given to pregnant females from the day of conception. Dox was removed on the day the pups were born. Offspring of Gt(ROSA)26Sor mice47 and double transgenic TgCN12; TgLC1 mice were used for detection of Cre activity. Stereotactic rAAV-Cre virus injection (Supplementary Figs. 1b and 2a) in Grin12lox mice was performed as described48.

In situ hybridization. In situ hybridizations were performed as described pre-viously14,49 on 15-µm cryostat sections. Antisense oligonucleotides (Grin1-nr1mouseis: 5′- GAA CTG ACA GCC CCA CCA GCA GCC ACA GTG TGC TC-3′ and Gria1-pan-A: 5′-GTC ACT GGT TGT CTG ATC TCG TCC TTC TTC AAA CTC TTC ACT GTG-3′) were 3′-end-labeled with terminal deoxynucleotidyl transferase and [α-33P]dATP.

Immunohistochemistry. Immunohistochemistry was performed as described14,50. Brain sections (50 and 100 µm) were stained with primary antibodies anti-Cre (PRB-106C, 1:8000, polyclonal, BabCO), anti-neuronal-specific nuclear protein (MAB377, NeuN; 1:100, monoclonal, Millipore) and anti-calbindin (code no. 300, 1:1000, monoclonal, Swant), glial fibrillary acidic protein (GFAP, ab16997; 1:100, Abcam) and anti-parvalbumin (P3088, 1:1000, monoclonal, Sigma). Secondary anti-mouse and anti-rabbit coupled to Cy3 and FITC were used (each 1:200; Jackson ImmunoResearch). Slices were mounted on slides, air-dried, covered with cover slips using either eu-kitt (O. Kindler) or Aqua-Poly/Mount (Polysciences). β-Galactosidase was visualized by incubating brain slices in X-gal14,49 in the dark at room temperature for 24 h. Subsequently, slices were washed twice in PBS, mounted and in some cases counterstained in eosin.

electrophysiological analysis of Grin1∆DGCA1 mice. Male and female mice (8–12 months old) were killed with Suprane (Baxter) and brains removed. Transverse slices (400 µm) were cut from the middle and dorsal portions of each hippocampus and prepared as described51. In one set of experiments, orthodromic stimuli (50 µs, <280 µA) were delivered alternately through two tungsten electrodes, one in stratum radiatum and another in stratum oriens of CA1. EPSPs were monitored by two glass electrodes (filled with ACSF) placed in the corresponding synaptic layers. After obtaining stable synaptic responses in both pathways (0.1 Hz for at least 10–15 min), the radiatum pathway was tetan-ized (1 s at 100 Hz, repeated four times at 5-min intervals), whereas the oriens pathway served as an untetanized control pathway. To standardize tetanization strength in different experiments, the tetanic stimulation strength was set in response to a single shock at intensity just above the threshold for generating a population spike.

In a second series of experiments, a stimulation electrode was placed in stratum radiatum at the border between CA1 and CA3. Stimulation at 0.1 Hz elicited synaptic responses recorded simultaneously by two glass electrodes localized to stratum radiatum in CA1 and CA3 (Fig. 2c). Pathways were tetan-ized as described above. Synaptic efficacy was assessed by measuring the slope of the fEPSP in the middle third of its rising phase. Six consecutive responses (1 min) were averaged and normalized to the mean value recorded 4–7 min before tetanic stimulation.

Subjects for behavioral studies. All mice were bred at the Max Planck Institute of Medical Research (Heidelberg, Germany). In Oxford (UK) mice were main-tained in groups, in a humidity- and temperature-controlled environment, under a 12-h light/dark cycle (lights on at 07:00). Testing took place during the light phase. All mouse lines were kept on the C57Bl/6N genetic background (Charles River) for 6–10 years before setting up breeding cohorts for this study. Male and female mice at least 4–5 months of age were used. There were no significant main effects of sex, or interactions involving sex and genotype. Therefore sex has been removed as a factor from the analyses reported. All behavioral testing was conducted by experimenters blind to genotypes. Cytotoxic hippocampal lesion animals (Supplementary Fig. 5) were generated as described52.

Spatial reference memory water maze task. Spatial reference memory was assessed during acquisition of the fixed location, hidden platform water maze task, as described22,53. Experimentally naive control (n = 12) and Grin1∆DGCA1 mice (n = 12) were compared at 4–5 months of age. Mice had no prior water maze pre-training.

Spatial reference memory radial maze task. Control (n = 12) and Grin1∆DGCA1 mice (n = 11) were compared during acquisition of the spatial reference memory radial maze task at 8–9 months of age, as described previously14. To specifically assess spatial reference memory, three of the six arms were baited with 0.1 ml condensed milk (Fig. 4). Each entry into an unbaited arm was scored as a refer-ence memory error (maximum of three errors per trial). Mice were only able to enter each arm once on any given trial (that is, it was not possible to make working memory errors)54.

Spatial reference memory Y-maze tasks. The appetitively motivated and the swim-escape Y-maze spatial reference memory tasks (Supplementary Fig. 6) were performed as described26,53.

Appetitive visual discrimination task. Appetitively motivated, visual discrimination learning was assessed in a T-maze with a gray start arm plus two removable goal arms, as described26. The walls and floor of one goal arm were painted with black and white stripes, while those of the other goal arm were dark gray. Controls (n = 13) and Grin1∆DGCA1 mice (n = 11) were required to associate a particular color of goal arm with reward.

Spatial discrimination beacon task. Experimentally naive controls (n = 11) and Grin1∆DGCA1 mice (n = 12), 8–9 months of age, were trained on a spatial dis-crimination version of the water maze task using two visibly, identical beacons. Mice were first trained to approach a single, black, spherical beacon (diameter 15 cm; 24 cm high), sitting on the water surface, which indicated the position of the escape platform (eight trials per day for 3 d). The spatial location of the beacon and the start position of the animal changed between trials, according to a pseudo-random sequence.

Mice were then trained to discriminate two identical, visible beacons, depend-ing on their spatial locations. Both beacons remained in fixed locations in space throughout testing and were in diametrically opposite quadrants of the pool. One beacon indicated the platform position (S+ beacon), whereas the other beacon was attached to a thin metal pole to hold it in a fixed position at the water’s surface, but provided no means of escape (S− beacon; Fig. 6a). The allocation of mice to particular platform locations was counterbalanced with respect to genotype. The identities of the actual physical beacons that were (i) placed on the platform and (ii) attached to the thin metal pole were changed every 24 trials.

Mice were placed into the water facing the sidewall from one of six possible start locations according to a pseudo-random sequence. For half of the trials the S+ beacon and platform were to the left of the start position and for half of the trials they were to the right. In addition, two start positions were equidistant between the beacons, two were closer to the S+ beacon/platform (~80 cm), and two were closer to the S− beacon, and thus further from the S+ beacon/platform (~140 cm; Fig. 6b). Mice received eight trials per day. Therefore, over 3 d of testing each mouse received four trials from each possible start position. They received no more than three consecutive trials from the same start position. In total, they received 15 d of training (five blocks; 24 trials per block).

On top of each beacon was a circular (20 cm diameter) piece of laminated white card. On the S+ beacon, this white circle sat exactly above the position of

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nature neurOSCIenCedoi:10.1038/nn.3166

the platform. Mice were considered to have made an error and chosen the wrong beacon when they passed under the white circle on the S− beacon. Whether the first choice that the mouse made was correct or incorrect was recorded (first-choice accuracy). In addition, we also counted the total number of errors made on a given trial. For example, if a mouse swam under the S− beacon and then re-emerged before again swimming under the S− beacon, then this was scored as two errors. In addition, two probe trials were conducted 24 h after the previous training trial. Both beacons and the platform were removed from the pool and the mouse allowed to swim freely for 60 s. A first probe test was conducted after 72 trials and a second probe test after 120 trials.

Visual discrimination beacon task. Experimentally naive controls (n = 8) and Grin1∆DGCA1 mice (n = 9), 8–9 months of age, were trained on a non-spatial ver-sion of the beacon task. Mice were required to discriminate between two visually distinct beacons (a gray funnel versus a black-and-white striped cylinder) to locate the hidden platform (20 cm diameter; Fig. 7). Mice first received 3 d of pre-training during which they were required to swim to a single black spherical beacon, in a variable spatial location, that indicated the position of the platform (eight trials per day for 3 d).

Mice were then required to choose between a gray funnel (17.5 cm high, 17.5 cm diameter at the top, 2 cm stem) and a black-and-white striped cylinder (17 cm high, 12 cm diameter at the top, 10 cm diameter at the bottom, 2 cm thick stripes) (Fig. 7a). One cue was always positioned on top of the platform, whereas the other cue was attached to a thin metal pole and provided no escape from the water. The allocation as to whether the funnel or the cylinder was associated with the platform was counterbalanced with respect to genotype.

The spatial location of the platform was varied pseudorandomly between the four quadrants of the pool. Therefore there was no spatial solution to the task. The incorrect (S−) beacon was always located in the diametrically opposite quad-rant. There were six possible start locations for each possible platform position (Fig. 7b). Mice were placed into the water facing the sidewall from one of the start locations according to a pseudorandom sequence. For half of the trials the S+ beacon/platform was to the left of the start position and for half of the trials it was to the right. In addition, two start positions were equidistant between the beacons, two were closer to the S+ beacon and platform, and two were closer to the S− beacon, thus mimicking the spatial discrimination task. Mice received eight trials per day. Therefore, over 3 d of testing each mouse received six trials

with the S+ beacon/escape platform in each of the four possible spatial locations, with one trial from each of the possible start positions for that given platform location. Mice received no more than three consecutive trials from the same start position. In total, they received 18 d of training (six blocks; 24 trials per block). Whether the first choice made by the mouse was correct or incorrect and the total numbers of errors per trial were recorded, as before.

Statistical analysis. Data from behavioral experiments were analyzed using repeated-measures ANOVAs. Subsequent post hoc investigations involved analysis of simple main effects and unpaired, two-tailed t-tests where appropriate. For analysis of time spent in each quadrant during water maze probe tests, because the fourth quadrant data point was never independent of the other three, P values were adjusted to reflect a reduction in the degrees of freedom in both the main effect of quadrant and the group × quadrant interaction. Data from electrophysi-ological experiments were analyzed using paired Student’s t test (two-tailed).

46. Lee, G. & Saito, I. Role of nucleotide sequences of loxP spacer region in Cre-mediated recombination. Gene 216, 55–65 (1998).

47. Soriano, P. Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat. Genet. 21, 70–71 (1999).

48. Tang, W. et al. Faithful expression of multiple proteins via 2A-peptide self-processing: a versatile and reliable method for manipulating brain circuits. J. Neurosci. 29, 8621–8629 (2009).

49. Jerecic, J. et al. Impaired NMDA receptor function in mouse olfactory bulb neurons by tetracycline-sensitive NR1 (N598R) expression. Brain Res. Mol. Brain Res. 94, 96–104 (2001).

50. Krestel, H.E., Mayford, M., Seeburg, P.H. & Sprengel, R. A GFP-equipped bidirectional expression module well suited for monitoring tetracycline-regulated gene expression in mouse. Nucleic Acids Res. 29, E39 (2001).

51. Jensen, V. et al. A juvenile form of postsynaptic hippocampal long-term potentiation in mice deficient for the AMPA receptor subunit GluR-A. J. Physiol. (Lond.) 553, 843–856 (2003).

52. Deacon, R.M., Bannerman, D.M., Kirby, B.P., Croucher, A. & Rawlins, J.N. Effects of cytotoxic hippocampal lesions in mice on a cognitive test battery. Behav. Brain Res. 133, 57–68 (2002).

53. Reisel, D. et al. Spatial memory dissociations in mice lacking GluR1. Nat. Neurosci. 5, 868–873 (2002).

54. Schmitt, W.B., Deacon, R.M., Seeburg, P.H., Rawlins, J.N. & Bannerman, D.M. A within-subjects, within-task demonstration of intact spatial reference memory and impaired spatial working memory in glutamate receptor-A-deficient mice. J. Neurosci. 23, 3953–3959 (2003).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.