Colin Lever Institute of Psychological Sciences University of Leeds

Colin LeverInstitute of Psychological SciencesUniversity of Leeds

ART PhD student Day, 15th March 2011

Studying cognitive processes in freely behaving rodents: neurons, oscillations, and behaviour

(focusing on hippocampal formation)

Plan of the talk

Why focus on the hippocampus? Which regions degenerate first in classic AD?

Outline characteristics of neurons supporting spatial cognition and memory in Hippocampal formation

Outline Theta oscillation-related changes in environmental novelty (encoding-related changes?)

THEN: 2 rodent AD models:

one with theta-related impairments,

one with CA1 place cell impairments

Why focus on the hippocampal formation?

Hippocampus has been linked to memory since H.M.’s devestating memory loss following removal of hippocampus & surrounding tissue

In animal literature, two key discoveries in the early 1970s:

1) LTP (Bliss and Lomo, 1973)

2) Place cells (O’Keefe and Dostrovsky, 1971)

The Hippocampus is the first region to degenerate in ‘classic’ Alzheimer’s dementia

Stages in Alzheimer’s disease:The spread from entorhinal cortex & CA1

Densities of Neurofibrillary tangles in mm2 in various brain regions amongst 7 groups defined by patterns of damage. These groups are then used ‘post hoc’ to predict clinical features.

Groups 1, 2, 3, 4, 5, 6, 7

Groups 1, 2, 3, 4, 5, 6, 7

Corder et al, 2000, Exp Gerontol

Stages in Alzheimer’s disease:The spread from entorhinal cortex & CA1

Group 1 = ‘normal aged’, Groups 2 & 3 = ‘possible AD’,

Group 4, 5, & 6 = ‘probable AD’

Group 7 = ‘definite AD’

Groups 1, 2, 3, 4, 5, 6, 7

Corder et al, 2000, Exp Gerontol

Layer II entorhinal cells are critical

Profound Loss of Layer II Entorhinal Cortex Neurons Occurs in Very Mild Alzheimer's Disease

Teresa Gómez-Isla, Joseph L. Price, Daniel W. McKeel Jr., John C. Morris, John H. Growdon, and Bradley T. Hyman

Journal of Neuroscience, 1996, 16: 4491-4500

‘A marked decrement of layer II neurons distinguishes even very mild AD from nondemented aging’.

Basic findings replicated by:Kordower et al, 2001, Annals of Neurology 49: 202-213

MCI and mild AD = fewer/atrophied Entorhinal layer II neurons


Kordower et al, 2001, Annals of Neurology 49: 202-213

No cog impairment

Mild cog impairment

Alzheimer’s disease

Layer 2 ‘islands’Layer 2 ‘islands’



No cog impairment

Mild cog impairment



Very few layer 2 neurons



No cog impairment

Mild cog impairment





Stages in Alzheimer’s disease:The spread from entorhinal cortex

No cognitive impairment -> Mild cognitive impairment ->Early stage AD -> Developed AD

Entorhinal cortex (esp. layer 2) ->CA1 ->Subiculum CA3 ->MTL and temporal cortex ->Other neocortex and subcortical regions

Where to focus in the hippocampal formation?

The Hippocampal formation (HF) is the first region to degenerate in ‘classic’ Alzheimer’s dementia

Regions affected early on:

Entorhinal cortex, CA1, Subiculum

The HF is part of ‘septo-hippocampal’ theta system. Medial Septum/DBB has an important role in controlling hippocampal theta.

So to develop useful rodent AD models, we need to establish normal physiology and function of neurons and oscillations in the rodent HF.

How can we go about doing that?

Extracellular recording in freely moving rodent

Histology confirms the recording sites

of the electrodes

Electrodes gradually lowered to target site over days/weeks

e.g. one site is CA1 pyramidal layer

Example configuration of 1 drive

e.g. other site is Hpc fissure

Multi-site dual-drive extracellular recording (64ch)



of the electrodes





Multi-site dual-drive extracellular recording (64ch) camera

Spikes& LFP

Track Head position & orientation: LEDs on front & back of head

HP

RecordingEnvironment(bird’s eye view)



of the electrodes






Spikes& LFP


10.1 peak rate (Hz)

Place cell Firing rate mapPlace cell Spike location plot



of the electrodes






Spikes& LFP


LFP showing theta oscillation

Am

plitu

de (m

V)

Time (seconds)

‘Raw’ theta (broad low-pass filter)

Analytic theta (apply offline 6-12 Hz filter, then Hilbert transform)

Dashed Lines indicate theta peak

Extracellular spike waveform on each of 4 tetrode tips

‘Place cells’ in CA1

Coloured square indicateswhere rat was when cell fired

Firing rate maps

(taking dwell timeinto account)

HP

Bird’s eye view of recording environment

All spikes Averaged spike

Extracellular recording in freely moving rodent:Recording many neurons simultaneously

What do neurons do in different hippocampal regions?

CA1 pyramidal cells are ‘place cells’.

Entorhinal cortex contains different types of spatial cells. Layer 2 cells are often ‘grid cells’.

Subiculum contains different types of spatial cells. Some act like place cells. Some are boundary vector cells. Some are grid cells.

We need to develop some idea of how neurons function normally, before we know how to look for impairment.

What do neurons do in region CA1?

CA1 pyramidal cells are ‘place cells’.

CA1 place cells show context-specific firing (later slides).

Simultaneously recorded CA1 place cells

A few cells cover the whole environmentThe active cells in that environment embody the ‘Cognitive Map’ of that environmentThey code for location AND spatial context

Lever et al, Nature, 2002

What do neurons do in entorhinal cortex?

Entorhinal cortex cells are heterogenous population:

Grid cells most striking discovery (Hafting et al, Nature, 2005). Many Layer II stellate cells are grid cells.

So this may be the first thing that goes wrong in human AD. And if a rat AD model could recapitulate human disease progression, you must understand grid cells.

Grid cells (found in Entorhinal Ctx, presubiculum, parasubiculum, and subiculum)

13.2 Hz

9.7

17.5

5.8

Large scaleLong distance between peaks~ 100 cm

Intermediate scale

Small scaleShort distance between peaks~30 cm

13.2 Hz

Mammalian brain divides the environment into triangular grids(broadly equilateral)

Each grid cell has a characteristic spatial scale

17.5

5.8


Intermediate scale


9.7

Grid cells (found in Entorhinal Ctx, presubiculum, parasubiculum, and subiculum)

Theta frequency & gain of movement-speed signal

13.2 HzSpatial scale related to systematic variation in the gain of a movement-speed signal (theta frequency changes)

Lower theta frequency MPOs in ventral Entorhinal grids, where grids have large spatial scale

Higher theta frequency MPOs in dorsal EC grids, where grids have small spatial scale

Grids seem to provide a strong spatial metric signal, encode distance travelled?

9.7

17.5

5.8

Grid cells


Intermediate scale


Burgess et al Hippocampus 2005

Code for Head Direction irrespective of location

e.g. the 4 quadrants of a cylinder

The brain’s compass

Parallel vectors

The four vectors do not converge on a point in the distance

Head direction cells (presubiculum, entorhinal ctx)

What do neurons do in Subiculum?

Subiculum contains different types of spatial cells.

Some act like place cells (shown).

Some are grid cells (shown)

Some are boundary vector cells (next slides).

Boundary Vector cells in the Subiculum

(Lever et al, 2009, Journal of Neuroscience)

What constitutes a boundary? Wall-less Environments

13.2 Hz

50-cm high walls

No walls (drop)

No walls (drop)10 cm gap between the 3 squares


13.2 Hz

50-cm high walls

No walls (drop)

No walls (drop)10 cm gaps between the 3 squares

Rat walks across drop


13.2 Hz

50-cm high walls

No walls (drop)


Rat walks across drop


13.2 Hz

50-cm high walls

No walls (drop)



13.2 HzSo Subicular boundary vector cells appear to function as high-level spatial perceptual cells

Wall and drop don’t share the same visual properties. And BVCs fire in darkness.

Function?Spatial Inputs to place cellsAnchor grids to external boundaries?

Are these cell types found in humans?

Yes, and if not, seems very probable.

Place cells: monkeys, humans (Ekstrom et al, Nature, 2003)

Head direction cells: in monkey presubiculum.

Grid cells: Indirect fMRI evidence (Doeller et al, Nature, 2010)

Boundary vector cells: not yet looked for (recent discovery)

Population signal of predicted grid cell activity in right entorhinal cortex

Strong links between spatial/context memory system in rats and autobiographical memory in humans

So if we understand the hippocampal system in rodents at the level of neurons and oscillations

we will be able to create more precise rodent AD models of episodic/autobiographical memory deficits

and provide a more accurate platform for testing therapeutic agents

Do hippocampal neurons show learning? What does it look like at the neuron level?

Contextual discrimination learning

Square vs Circle


Slow Contextual discrimination learning:Can we observe learning develop over time?Can we see memory after a delay?

Incidental learning paradigm:Experimenter does nothing to encourage the discrimination learning


Slow Contextual discrimination learning:

Quite a hard task for the rat?Like too-similar floors in car park? – Takes a while to discriminate.

4.43.6 2.8

5.12.6 2.1

0.33.3

4.03.11.1

8.4 0.1 0.3 0.00.2

3.1 2.9 8.12.6 5.4 1.5 2.1

6.2 0.5 0.01.0 0.2 0.2 0.6

0.63.2

3.1 1.7

3.9 0.2 0.7 0.0

2.3

1.0

5.32.0

1 2 3

1 2 54 6 7 8

109 1 2 3 5

1 2 3 4 5

D1

D3

D7

D5

Fields initially similar

Contextual discrimination in place cells

4.43.6 2.8

5.12.6 2.1

0.33.3

4.03.11.1

8.4 0.1 0.3 0.00.2

3.1 2.9 8.12.6 5.4 1.5 2.1

6.2 0.5 0.01.0 0.2 0.2 0.6

0.63.2

3.1 1.7

3.9 0.2 0.7 0.0

2.3

1.0

5.32.0

1 2 3

1 2 54 6 7 8

109 1 2 3 5

1 2 3 4 5

D1

D3

D7

D5

Lever, Wills, Cacucci, Burgess, O’Keefe, Nature, 2002

Fields initially similar, then over time cells develop discriminatory firing (slow remapping)


4.43.6 2.8

5.12.6 2.1

0.33.3

4.03.11.1

8.4 0.1 0.3 0.00.2

3.1 2.9 8.12.6 5.4 1.5 2.1

6.2 0.5 0.01.0 0.2 0.2 0.6

0.63.2

3.1 1.7

3.9 0.2 0.7 0.0

2.3

1.0

5.32.0

1 2 3

1 2 54 6 7 8

109 1 2 3 5

1 2 3 4 5

D1

D3

D7

D5


Fields initially similar, then over time cells develop discriminatory firing (slow remapping):Cell fires in one environment, but not in another


4.43.6 2.8

5.12.6 2.1

0.33.3

4.03.11.1

8.4 0.1 0.3 0.00.2

3.1 2.9 8.12.6 5.4 1.5 2.1

6.2 0.5 0.01.0 0.2 0.2 0.6

0.63.2

3.1 1.7

3.9 0.2 0.7 0.0

2.3

1.0

5.32.0

1 2 3

1 2 54 6 7 8

109 1 2 3 5

1 2 3 4 5

D1

D3

D7

D5


Fields initially similar, then over time cells develop discriminatory firing (slow remapping):Cell fires in one environment, but not in another, orCell fires in different locations in each environment (less common)


4.43.6 2.8

5.12.6 2.1

0.33.3

4.03.11.1

8.4 0.1 0.3 0.00.2

3.1 2.9 8.12.6 5.4 1.5 2.1

6.2 0.5 0.01.0 0.2 0.2 0.6

0.63.2

3.1 1.7

3.9 0.2 0.7 0.0

2.3

1.0

5.32.0

1 2 3

1 2 54 6 7 8

109 1 2 3 5

1 2 3 4 5

D1

D3

D7

D5


Fields initially similar, then over time cells develop discriminatory firing (slow remapping)Day 1: 3/3 similarDay 3: 2/7 similarDay 5: 1/7 similarDay 7: 0/5 similarObserve development of learning!


D a y 1 :

S e r i e s

s t a r t

D a y 2 1 :

S e r i e s

E n d

D a y 7 1 :

2 n d D e l a y

t e s t

D a y 1 :

F i r s t

E x p o s u r e s

S e r i e s

E n d

D a y 2 1 :

1 7 d a y s

1 s t D e l a y

t e s t

2 8 d a y s

2 n d D e l a y

t e s t

D a y 7 1 :


Long-term memory

Representations initially similar

Over time, cells learn to discriminate the 2 shapes

Memory for what has been learned?

D a y 1 :

S e r i e s

s t a r t

D a y 2 1 :

S e r i e s

E n d

D a y 7 1 :

2 n d D e l a y

t e s t

D a y 1 :

F i r s t

E x p o s u r e s

S e r i e s

E n d

D a y 2 1 :

1 7 d a y s

1 s t D e l a y

t e s t

2 8 d a y s

2 n d D e l a y

t e s t

D a y 7 1 :


Long-term memory

Representations initially similar

Over time, cells learn to discriminate the 2 shapes

Memory for what has been learned? YES!

Summary: CA1 neurons ‘learn’ to discriminateIndividual CA1 neurons show ‘long-term plasticity’

Discrimination is observed to increase with more experience of contexts

Once learned, the discrimination is remembered after month-long delay

Intentionally very different spatial contexts

Days 1 to 5 Day 6, 8, 10 Standard Altered (3rd, 4th)

Both walled environments:

D oor

H P

Rec

ordi

ngsy

stem

B lack C urtains

C ue card

D oo r

H P

Rec

ordi

ngsy

stem

H P

Shelves

C ue card

Shelves

Holding platform

1st

2nd

4th

6th

5th

3rd

Trial Sequence Environment

Context-specific firing can develop rapidly if contexts are significantly different

Rat 1 Rat 2 Rat 3 Cell 1 Cell 2 Cell 1 Cell 2 Cell 1 Cell 2

13 Hz 2 Hz 3 Hz 2 Hz 6 Hz

16 2 6 3 13

5 2 4 2

8 4 6 3

12 5 3 7 9

3 5 5 16 6

In this experiment, place cells have ‘remapped’ the different contexts already within the 10-15 minute total trial time in each context


Lever et al, unpublished data

Rat 1 Rat 2 Rat 3 Cell 1 Cell 2 Cell 1 Cell 2 Cell 1 Cell 2

13 Hz 2 Hz 3 Hz 2 Hz 6 Hz

16 2 6 3 13

5 2 4 2

8 4 6 3

12 5 3 7 9

3 5 5 16 6

As with slow discrimination for subtly-differing context, a) a place cell can discriminate by firing in one context but not another, or by firing in both contexts but in different locations b) it’s incidental learning


The hippocampal theta oscillation is sensitive to novel contexts

Theta Phase and Memory statesHippocampal LTP protocols are optimal using stimulation at theta frequencyTheta phase determines whether LTP is achieved, e.g. in CA1 stimulate at theta peak -> strongest LTP

LTPWell-establishedresult

LTD or no change results

Model (Hasselmo et al, 2002) links these plasticity results to memory states. In novelty-elicited encoding there should be:

a bias -> information from entorhinal cortex, presumed to arrive near peak of principal-cell layer theta

Vs in retrieval, a bias -> predictive CA3 input (arriving at trough)

Every spike is assigned a theta phase of firing

We then aggregate all the spikes’ theta phases from:a) CA1 b) Subiculum

Later CA1 mean theta phase in novelty

Each polar plot represents all recorded CA1 spikes in that trial.Mean spike phase normalised such that mean phase of all CA1 spikes in last trial in familiar environment (‘Baseline’) is 0°.

k = circular concentration m = mean phase

Highly familiarenvironment

Very different Novelenvironment

Conclusion:Theta phase may separate encoding and retrieval

If we can assume:

More Encoding during Novelty trials than in Familiar trials

Then our results suggest that theta phase could play a role in plasticity in the hippocampal memory system, and the balance between encoding and retrieval

Likely a general coding strategy in the brain?

Novel environments elicit theta frequency reduction


Decrease in theta frequency of up to 1 Hz recorded in each rat in the novel environment.


NovelEnvts.

Familiar Envt.

Novel environments elicit theta frequency reduction:Summary

Jeewajee, Lever et al (2008) Hippocampus

Summary: Hippocampal theta and novelty

Novel environments elicit:1) Later theta phase of firing in CA1 neurons (Lever et al,

2010, Hippocampus)

2) Lower theta frequency in hippocampal theta (Jeewajee, Lever et al, 2008, Hippocampus)

This second finding is (relatively) easy to study.

This could be explored in rodent AD models without needing to record hippocampal neurons.

Decreased rhythmic GABAergic septal activity & memory-associated theta oscillations after hippocampal Amyloid-b pathology in the rat

Basic idea: a) Inject long-lasting Ab aggregates (Ab40 & Ab42 in 2:1 ratio) bilaterally into

4 injection sites in the dorsal hippocampus. [Ab40 20 mg/ml & Ab42 10 mg/ml, Bachem, 0.25 ml per injection site]

b) Implant electrodes to record local field potentials from the hippocampus (a little posterior to injection sites)

c) Give rats recognition memory task every two days for 3 weeks (first formal test one day after injection), evaluate progressive impairment

d) Test theta power over course of experimente) Detailed analysis of theta oscillations and behaviour on key days

(D1, D7, D15, D21)

Villette et al (2010) J Neurosci


New StimuliEmpty Position

Long termNo change

Ab rats show similar investigativerepertoire to controls


Ab rats overexplore the familiar items, & underexplore the novel items


Long termNo change

Classic memory test in rodents. Rats should explore new/changed items more.Authors used rats’ investigative rearing.

Investigative behaviour is not selectivelyincreased for the new/changed items in Ab rats.

I.e. Ab rats show memory deficit

What about neurophysiological correlates?



Ab rats develop reduced theta powerAb rats overexplore the familiar items, & underexplore the novel items



Long termNo change


The reduced theta power Ab rats develop is non-specific.

It occurs regardless of the task and old/new space/object combinations.e.g. Tested different group of Ab rats and controls who are exposed to unchanging stimuli in context. These Ab rats also show reduced power.Is there a neural correlate specific to the old/new memory impairment?

Ab rats overexplore the familiar items, & underexplore the novel items



Long termNo change

“Loss of task-related theta frequency modulation after hippocampal Ab injection” Villette et al (2010)

J Neurosci

Ab rats

Controls

On Days 15 & 21, control rats show behavioural discrimination of old vs new items. Ab rats don’t. Thus, in parallel with memory deficits, Ab rats do not show the novelty-elicited theta frequency reduction which emerges in controls by D15 & D21.

Ab rats show reduced theta power

Ab rats do NOT show newvs old theta frequency difference


Villette et al studied spatial/object associational novelty. They replicate in their controls the Jeewajee, Lever et al (2008) result based on environmental novelty:

New spatial/object combinations elicit higher levels of investigation and lower-frequency theta oscillations in controls.

Neither occurs in rats injected with Ab aggregates

Discovering neurophysiological correlates of spatial/contextual representation and memory are useful in building more precise animal models of dementia

That can provide a bridge between molecules and behaviour.

Villette et al (2010) J Neurosci

Place cells can provide an intermediate level of investigation between

molecules and behaviour

Research goals: • study the network properties of hippocampal cells

in rodent models of Alzheimer’s disease.• investigate relationships between physiological

and cognitive changes during the progression of the disease.

One experimental model: the Tg2576 mouse as a model of

‘Alzheimer-like’ dysfunction

• neuronal overexpression of a mutated form of human amyloid (APP695SWE).

• develops elevated brain levels of soluble amyloid by 6-8 months, and neuritic plaques by 10-16 months.

• age-dependent impairment on spatial navigation/memory tasks.

Lab Setup

Young mice: performance at different delays

1) Behaviour

2) HPC place cells

Aged mice: performance at different delays

Delay p < 0.001

Genotype p < 0.005

Place cells in aged mice

Quantifying Spatial Characteristics of

the Place Fields

Correlation between behaviour and Spatial information

Basic Physiological Properties

Conclusions• Place cell signalling is normal in young tg2576

mice but disrupted in some aged tg2576 mice.• There is a correlation between place cell

disruption and spatial memory deficits.• Combining place cell recording with spatial

memory testing will provide a powerful tool for investigating molecular changes which lead to the physiological alterations in Alzheimer’s disease and for testing possible therapeutic strategies.

Overall conclusion

Neurophysiology in behaving rodents linking neurons and oscillations to behaviour

Is a useful and arguably necessary step

In creating good AD models in rodents

Thanks to: LEEDS: Christine Wells, Ali Jeewajee, Sarah Stewart, Vincent Douchamps,

UCL: Ali Jeewajee, Stephen Burton, Francesca Cacucci, Tom WillsNeil Burgess, John O’Keefe

And you for listening!

End

Documents

Colin Lever Institute of Psychological Sciences University of Leeds