Cholinergic circuit modulation through differential...

J Physiol 592.19 (2014) pp 4155–4164 4155

The

Jou

rnal

of

Phys

iolo

gy

SYMPOS IUM REV IEW

Cholinergic circuit modulation through differentialrecruitment of neocortical interneuron types duringbehaviour

Rogier B. Poorthuis, Leona Enke and Johannes J. Letzkus

Max Planck Institute for Brain Research, 60438 Frankfurt, Germany

Abstract Acetylcholine is a crucial neuromodulator for attention, learning and memory. Releaseof acetylcholine in primary sensory cortex enhances processing of sensory stimuli, and manyin vitro studies have pinpointed cellular mechanisms that could mediate this effect. In contrast,how cholinergic modulation shapes the function of intact circuits during behaviour is onlybeginning to emerge. Here we review recent data on the recruitment of identified interneurontypes in neocortex by cholinergic signalling, obtained with a combination of genetic targetingof cell types, two-photon imaging and optogenetics. These results suggest that acetylcholinerelease during basal forebrain stimulation, and during physiological recruitment of the basalforebrain, can strongly and rapidly influence the firing of neocortical interneurons. In contrastto the traditional view of neuromodulation as a relatively slow process, cholinergic signallingcan thus rapidly convey time-locked information to neocortex about the behavioural state of theanimal and the occurrence of salient sensory stimuli. Importantly, these effects strongly dependon interneuron type, and different interneuron types in turn control distinct aspects of circuitfunction. One prominent effect of phasic acetylcholine release is disinhibition of pyramidalneurons, which can facilitate sensory processing and associative learning.

(Received 4 March 2014; accepted after revision 19 May 2014; first published online 30 May 2014)Corresponding author J. J. Letzkus: Max Planck Institute for Brain Research, 60438 Frankfurt, Germany.Email: [email protected]

Abbreviations PV, parvalbumin; SOM, somatostatin; VIP, vasoactive intestinal polypeptide.

Introduction

The function of neuronal circuits is flexible andcontinuously adjusted to current behaviouralrequirements (Bargmann, 2012), a process in which

Rogier B. Poorthuis obtained his PhD from the VU University Amsterdam, where he investigatedmodulation of prefrontal cortical circuits by nicotinic acetylcholine receptors and nicotine, and sub-sequently joined the Letzkus lab. During his PhD at the John Curtin School of Medical Research(Canberra, Australia). Johannes J. Letzkus examined dendritic and axonal mechanisms of neocorticalinformation processing. As a postdoc and subsequently an independent fellow at the Friedrich MiescherInstitute for Biomedical Research (Basel, Switzerland), he started to analyze the circuit mechanismsof associative learning in cortex. In 2013 he moved to the Max Planck Institute for Brain Research(Frankfurt, Germany), where he is currently an independent group leader. The lab focuses on themechanisms and consequences of neocortical circuit modulation by behavioral context, combiningcell-type specific recording and activity perturbation approaches with learning and attention tasks.

R. B. Poorthuis and L. Enke contributed equally to this work.

This review was presented at the symposium Synaptic properties and functional consequences of cholinergic transmission in the CNS, which took placeat the annual meeting of the Society for Neuroscience, San Diego, CA, USA on 10 November 2013.

neuromodulators are critically involved (Lee & Dan,2012). Acetylcholine is an essential neuromodulator forthe coordination of behavioural state, arousal, attentionand learning and memory (Sarter & Bruno, 1997). Themain source of neocortical acetylcholine derives from

C© 2014 The Authors. The Journal of Physiology C© 2014 The Physiological Society DOI: 10.1113/jphysiol.2014.273862

4156 R. B. Poorthuis and others J Physiol 592.19

the basal forebrain, which sends diffuse projectionsthroughout the cortical mantle (Woolf, 1991). Acetyl-choline release from these projections over long timescales(minutes to hours) facilitates attentional processes anddetection of rare stimuli (Parikh et al. 2007; Sarteret al. 2009; Paolone et al. 2012). On shorter timescalesbasal forebrain neurons are activated upon presentationof novel sensory information (Wilson & Rolls, 1990b;Miranda et al. 2000) and encode the salience of sensorystimuli (Richardson & DeLong, 1990; Wilson & Rolls,1990a). In sensory cortex, acetylcholine enhances neuro-nal responses to relevant stimuli, decreases trial-to-trialvariability and induces gamma oscillations (Reynolds &Chelazzi, 2004; Herrero et al. 2008; Fries, 2009; Goard &Dan, 2009; Harris & Thiele, 2011). However, how theseeffects are produced mechanistically in the local circuit isonly beginning to emerge.

Cortical circuits are composed of local inhibitoryinterneurons and excitatory pyramidal cells, whichproject to long-range target areas. The effects of acetyl-choline on pyramidal neurons have been investigatedin great detail, and include direct depolarization andreduction of cortico-cortical input by muscarinic acetyl-choline receptor activation, as well as enhancementof thalamo-cortical input via nicotinic receptors(McCormick & Prince, 1985; Gil et al. 1997; Kimura &Baughman, 1997; Disney et al. 2007; Kawai et al. 2007).The combination of these effects is thought to increasethe signal-to-noise ratio for feed-forward thalamic input,thereby enhancing sensory processing during learning andattention (Hasselmo & Giocomo, 2006).

In addition to these direct effects of acetylcholine,activity of pyramidal neurons is also under tight controlof a diverse set of inhibitory interneurons. Differentinterneuron types control distinct aspects of pyramidalcell function, and thereby exert control over a broadrepertoire of circuit functions (Isaacson & Scanziani,2011; Kepecs & Fishell, 2014). Importantly, studies inbrain slices have shown that different interneuron typesalso display differential responses to acetylcholine (Bacciet al. 2005). Fast-spiking parvalbumin-positive (PV) inter-neurons targeting the perisomatic region of pyramidalcells for the most part do not display cholinergic responses(Kawaguchi, 1997; Xiang et al. 1998; Gulledge et al. 2007;Kruglikov & Rudy, 2008). In contrast, dendrite-targetinginterneurons identified by somatostatin (SOM) expressioncan be depolarized by activation of muscarinic receptors(Porter et al. 1999; Fanselow et al. 2008). In cortical layers2–6, a group of interneurons expressing 5HT3a receptors isdepolarized by acetylcholine acting on nicotinic receptors(Gulledge et al. 2007; Lee et al. 2010; Poorthuis et al. 2013).A subpopulation of these neurons co-expresses vasoactiveintestinal polypeptide (VIP; Porter et al. 1999; Lee et al.2010) and preferentially targets other interneurons (Davidet al. 2007; Pfeffer et al. 2013; Pi et al. 2013). Finally, all

interneurons located in cortical layer 1 are depolarizedfrom rest by nicotinic receptors (Christophe et al. 2002;Letzkus et al. 2011; Arroyo et al. 2012). This changes duringongoing firing, when acetylcholine leads to an inhibitionin layer 1 neurogliaform cells that in turn target pyramidalneurons (Brombas et al. 2014). In contrast, single-bouquetcells which preferentially target deeper layer interneurons(Jiang et al. 2013) display activation by acetylcholine underthese conditions.

These in vitro studies suggest that differentialmodulation of interneuron types may be an importantmechanism by which acetylcholine affects corticalcomputation. While the sparseness and heterogeneityof interneurons has for a long time hampered theirdetailed investigation in vivo, recently developed geneticapproaches for cell-type targeting (Taniguchi et al. 2011),combined with two-photon imaging (Kerr & Denk, 2008)and optogenetics (Zhang et al. 2007) now provide therequisite tools for these experiments. Here, we reviewthe available recent findings on how different types ofneocortical interneurons are recruited by the cholinergicbasal forebrain in vivo, and on how these effects combineto affect information processing in pyramidal neuronsduring behaviour.

Differential recruitment of neocortical interneurontypes by basal forebrain stimulation

To address the effects of acetylcholine release on differentgenetically identified neuron types, Alitto & Dan (2013)combined electrical stimulation of the basal forebrainwith two-photon calcium imaging in the superficiallayers of visual cortex in anaesthetized mice (Fig. 1A).This approach revealed time-locked responses in a sub-set of neurons, suggesting that acetylcholine is able torapidly modulate the firing rate of cells in visual cortex(Fig. 1B–C). Importantly, both the incidence and thepolarity of these responses depended on cell-type (Fig. 1Dand E). Only a small fraction of pyramidal neuronsdisplayed activation via muscarinic receptors in this study,which is likely to be a direct action of acetylcholine on thesecells (cf. McCormick & Prince, 1985; Gulledge et al. 2007).In contrast, the vast majority of both VIP and layer 1 inter-neurons were strongly activated by acetylcholine (Fig. 1D).In agreement with previous observations, layer 1 inter-neuron responses were abolished by nicotinic receptorantagonists (Fig. 1E; Christophe et al. 2002; Letzkus et al.2011), while VIP interneurons showed a mixed activationprofile that was partially blocked by both nicotinic andmuscarinic receptor antagonists (Kawaguchi, 1997; Porteret al. 1999). Importantly, both single-bouquet cells in layer1 and VIP interneurons preferentially contact other inter-neurons (Christophe et al. 2002; David et al. 2007; Letzkuset al. 2011; Jiang et al. 2013; Pfeffer et al. 2013; Pi et al.

C© 2014 The Authors. The Journal of Physiology C© 2014 The Physiological Society

J Physiol 592.19 Cholinergic circuit modulation via interneurons 4157

2013), suggesting that the net effect of their activationby acetylcholine is a reduction of ongoing inhibitionin the circuit. Consistent with this notion, the majorityof PV interneurons that responded to basal forebrainstimulation were inhibited through a combination ofnicotinic and muscarinic effects (Fig. 1E). Since thesecells for the most part do not display direct responsesto acetylcholine (Kawaguchi, 1997; Gulledge et al. 2007;Kruglikov & Rudy, 2008), the most parsimonious inter-pretation is indirect inhibition via single-bouquet cells inlayer 1 and VIP interneurons (cf. Arroyo et al. 2012). Asecond response type in PV interneurons was excitation,which was abolished by muscarinic antagonists and waslikely to be due to an increased drive from pyramidalneurons. Interestingly, excitation of PV interneurons byacetylcholine was converted to inhibition under blockof muscarinic receptors, and inhibition of PV cells wasconverted to excitation when nicotinic receptors wereblocked, suggesting that excitation and inhibition compete

in the same population of PV interneurons (Fig. 1E). Incontrast to these strong effects, only very few SOM inter-neurons displayed excitation or inhibition in response tobasal forebrain stimulation. Based on recent connectivitydata (Pfeffer et al. 2013; Pi et al. 2013), activation ofVIP interneurons would be expected to cause inhibitionof SOM cells. However, since calcium imaging pre-dominantly captures suprathreshold effects (Kerr & Denk,2008), it is likely that the low firing rate of SOM inter-neurons under these conditions may have preventeddetection of this inhibition. Indeed, in awake conditionsVIP interneurons have been found to inhibit SOM inter-neurons, a mechanism that is absent under anaesthesia(Lee et al. 2013; Fu et al. 2014).

In conclusion, these data demonstrate that neocorticalacetylcholine release elicits rapid, reliable and distinctresponses in several types of interneurons. Given thatdifferent interneuron types control distinct aspects ofpyramidal cell function, this is likely to be an important

PPV

SOMVIP

PV

Circuit BF stimulation BF stimulation+ Nicotinic receptor block

BF stimulation+ Muscarinic receptor block

L1 L1 L1 L1

Level of excitation Level of inhibition

L1

P

A B D

Ea EdEcEb

BFStimulation

Calcium Imaging

1

2

3 4

10%

Res

pons

e (d

F/F)

Time20 s 5 s

5 s

5 s

5 s6%

6%

1.5%

1.5%4

3

2

1

High HighLow Low

C

Negative

Positive

0

50

100

% R

espo

nsiv

e C

ells

50VIP PV SOMExc L1

L2/3 L2/3 L2/3L2/3

Figure 1. Modulation of genetically identified interneurons in visual cortex by basal forebrainstimulationA, schematic illustration of experimental design. B, two-photon calcium imaging of neurons in superficial layersof visual cortex loaded with OGB-1. C, dF/F traces of four example cells (indicated by numbers in B) during basalforebrain stimulation (arrows). C, right panels: average response over ten trials. Grey shading: 4× SD of baseline.D, percentage of neurons that displayed significant modulation by basal forebrain stimulation for different neurontypes. Note the strong excitatory response of layer 1 and VIP interneurons. PV interneurons were bidirectionallymodulated. Ea, schematic representation of genetically identified neurons under investigation. Eb–d, modulationof neuronal activity during basal forebrain stimulation in control (b) and during muscarinic (c) and nicotinic (d)receptor blockage. Colours indicate activation (red) and inhibition (blue). Neuron types with a low percentageof responsive cells are shown in grey. BF, basal forebrain; P, pyramidal neuron; PV, parvalbumin; VIP, vasoactiveintestinal polypeptide; SOM, somatostatin. Adapted with permission from Alitto & Dan, 2013.



and computationally rich mechanism by which acetyl-choline modulates information processing in neocorticalcircuits.

Cholinergic recruitment of neocortical interneurontypes during locomotion

While these data suggest that acetylcholine can exertdiverse and cell-type specific control over cortical circuits,only experiments in relation to behavioural functionscan directly address whether and when these mechanismsare recruited during physiological activation of the basalforebrain. One factor that strongly influences sensoryprocessing is the behavioural state of the animal, whichspans a wide range from sleep and quiet wakefulnessto active episodes (Niell & Stryker, 2010; Gentet et al.2012). Recent evidence suggests that locomotion stronglyenhances sensory responses in visual cortex relative toquiet wakefulness (Niell & Stryker, 2010), an effectaccompanied by changes in the excitation–inhibitionbalance (Bennett et al. 2013). Fu et al. (2014) investigatedthe circuit mechanisms underlying this effect usingtwo-photon calcium imaging of genetically identifiedinterneurons in the superficial layers of visual cortexwhile the mouse was free to run on a trackball (Fig. 2A).They observed that running episodes caused a markedand sustained increase in VIP interneuron activity(Fig. 2B), whereas unidentified neurons displayed no suchcorrelation. Interestingly, VIP interneurons in barrel andauditory cortex displayed a similar albeit less pronouncedmodulation. Blockade of nicotinic receptors stronglyattenuated the increase in VIP interneuron activity invisual cortex during locomotion (cf. Porter et al. 1999),while block of AMPA receptors had no effect (Fig. 2Cand D), indicating that the basal forebrain is a majorsource of locomotion-induced alterations in corticalstate. Consistent with strong connectivity between VIPand SOM interneurons (Pfeffer et al. 2013; Pi et al.2013), running was associated with a reduction in SOMinterneuron activity, whereas PV interneurons displayedheterogeneous modulation. These results suggest that oneprominent effect of locomotion could be disinhibition ofpyramidal neurons.

To directly address how VIP interneuron activationinfluences visual processing, the channelrhodopsin variantChETA (Gunaydin et al. 2010) was introduced into thesecells. In line with a disinhibitory role of VIP interneurons,responses of unidentified neurons to visual stimulationwere enhanced during VIP interneuron activation instationary animals (Fig. 2E and F). Moreover, optogeneticactivation of VIP interneurons caused no change inorientation selectivity, similar to the effect of locomotion(Niell & Stryker, 2010). Ablation of local VIP interneurons,on the other hand, strongly attenuated circuit modulation

by locomotion, indicating that VIP interneurons are bothnecessary and sufficient for regulating the gain of visualresponses during running (Fu et al. 2014).

These data demonstrate that acetylcholine releaseduring running rapidly recruits VIP interneurons toincrease the gain of sensory responses in pyramidalneurons through disinhibition, which in turn mightunderlie enhanced behavioural performance in a visualdetection task during locomotion (Bennett et al. 2013).This interpretation is also in line with a recent studyshowing that rapid and local enhancement of cholinergicsignalling by stimulating basal forebrain fibres in the visualcortex improves visual perception (Pinto et al. 2013).

Cholinergic recruitment of neocortical interneurontypes during learning

In addition to behavioural state, several forms ofassociative memory critically depend on the cholinergicsystem (e.g. Hasselmo, 2006; Weinberger, 2007). Recentresearch has addressed the circuit mechanisms of fearlearning in auditory cortex (Letzkus et al. 2011). Auditoryfear conditioning is a simple form of associative learningacquired by temporal coincidence of a neutral auditorystimulus with a mild foot-shock (LeDoux, 2000). Tounderstand the mechanisms mediating convergence ofthese two stimuli, Letzkus and colleagues asked whetherthe foot-shock alone elicits responses in auditory cortex(Fig. 3). Two-photon calcium imaging revealed that while,on average, foot-shocks caused no response in the neuro-nal population in layer 2/3, many layer 1 interneuronswere strongly activated. Targeted cell-attached recordingsshowed that foot-shocks elicit a strong increase in firingin approximately 75% of layer 1 interneurons, whereas theremainder displayed long-lasting inhibition (Fig. 3A–C).Surprisingly, the excitatory foot-shock response wasvirtually independent of glutamatergic transmission, butcompletely blocked by nicotinic antagonists (Fig. 3D). Inagreement with this pharmacology, microstimulation ofthe cholinergic basal forebrain evoked strong excitationof layer 1 interneurons similar to foot-shocks (Fig. 3E).Finally, excitatory foot-shock responses were also observedin layer 1 interneurons in primary visual cortex (cf. Alitto& Dan, 2013), a finding paralleled by the observation thatlocomotion leads to cholinergic activation of VIP inter-neurons in several cortical areas (see above, Fu et al. 2014),and consistent with the diffuse nature of projections fromthe basal forebrain. These data indicate that foot-shocksrapidly recruit the basal forebrain to activate a largepercentage of neocortical layer 1 interneurons throughnicotinic receptors. The latency of cholinergic activation(50–60 ms for foot-shocks; 10–20 ms for basal forebrainmicrostimulation) approaches the speed of conventionalsynaptic transmission, consistent with recent experiments



BA

D

Time (s)

Run

ning

spee

d (c

m/s

)R

espo

nse

am

plitu

de (Δ

F/F

%)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Bar direction (degree)

Res

pons

e (H

z)

No

LED

Pea

k R

espo

nse

C

With

LE

D P

eak

Res

pons

e

E

0

0.1

0.2

0.3

0.4

0.5

0.6

Zero

-tim

e cr

oss-

corr

elat

ion

**

F

n.s.

100

100

200

200

300

300

400

4000

4020

60

01020

0 10 20−10−20

0

−0.1

0.1

0.2

0.3

0.4

0.5

0.6

Cro

ss-c

orre

latio

n

B di ti (d )

Res

pons

e (H

z)

100 200 300 4000−1

5

4

6

0

2

1

3

7

Loading bufferMEC & MLANBQX

VIP neuron

Locomotion

No

ChE

TA

With

ChE

TAN

BQ

X

ME

C &

M

LA

Load

ing

buffe

r

With ChETA

Figure 2. Cholinergic recruitment of VIP interneurons during locomotionA, diagram displaying the experimental setup. A head-fixed mouse is free to run on a Styrofoam ball floating onair, while two-photon calcium imaging is performed in primary visual cortex. B, example traces showing runningspeed of the animal (bottom) and correlated activity of a VIP interneuron (top). C, cross-correlation (mean ± SEM)between running speed and the calcium response of VIP interneurons in control (red), and after blockade ofnicotinic (orange) and AMPA receptors (blue). D, the zero-time cross-correlation (mean ± SEM) between runningand VIP interneuron activity is strongly reduced by block of nicotinic acetylcholine receptors, but not by blockof glutamatergic receptors. E, orientation tuning of an unidentified unit in control (No LED, green), and duringoptogenetic activation of VIP interneurons using ChETA (Gunaydin et al. 2010; With LED, blue). Responses wereaveraged over five presentations of moving bars. F, average light-induced modulation of sensory responses incontrol mice (No ChETA), and in mice expressing ChETA in VIP interneurons. Adapted with permission from Fuet al. 2014.



AMPA block

GAD67-GFP

2P targeted recording

AP

freq

uenc

y [H

z]A

P fr

eque

ncy

[Hz]

AP

freq

uenc

y [H

z]

0

1

2

3

4

5

6

7

0

2

4

6

10

8

12

0 2 4 6−2

Time [s]

0

2

4

6

8

0 2 4 6−2

0 2 4 6−20

2

4

6

8

10

12

IPS

C fr

eque

ncy

[Hz]

0

5

10

15

20

25

0 2 4 6−2Time [s]

Control

Nic block

0

2

4

6

8

10

12

0 2 4 6−2

Sweeps

5%∆F/F

2 s

Sweeps + shocksSweeps Sweeps

+ shocks

0

200

400

600

∆F/F

are

a

***

PV

L1

L2/3

L1

P

L1

25 μm

2+

A Ca

F

B

ECb

D

Cc

Cd

Foot shock

0 2 4 6−2

1 s

2P Caimaging

BF stimulation

Figure 3. Foot-shock modulation of auditory cortex circuitsA, schematic illustration of experimental setup. Foot-shock (yellow) responses are measured using two-photontargeted loose-seal cell-attached recordings. B, schematic of different cell types under investigation. Colour codecorresponds to data in panel C. Ca, population response of layer 1 interneurons that are excited during foot-shocks(77%). Cb, inhibition of a smaller group of layer 1 interneurons (23%). Cc, firing of PV interneurons in layer2/3 is inhibited by foot-shocks. Cd, whole-cell recordings of pyramidal neurons reveal a reduction of inhibitorypostsynaptic currents (IPSCs) during foot-shock presentation. D, local block of nicotinic receptors (grey) stronglyreduces foot-shock responses in layer 1 interneurons, while local block of AMPA receptors (black) leaves theresponse intact. E, foot-shock activation of layer 1 interneurons is mimicked by basal forebrain stimulation. F,two-photon population imaging shows that foot-shocks significantly boost sensory responses to tone presentationin layer 2/3 of auditory cortex. Adapted with permission from Letzkus et al. 2011.



employing optogenetic stimulation of basal forebrainaxons in vitro (Bennett et al. 2012). Thus, on top of itswell-established slow mode of action, acetylcholine canalso convey rapid and time-locked information on theoccurrence of salient sensory stimuli to neocortex (Parikhet al. 2007; Sarter et al. 2009; Paolone et al. 2012).

How do cholinergic responses in layer 1 interneurons inturn affect processing of sensory information in auditorycortex? Single bouquet cells in layer 1 preferentially targetdeeper layer interneurons, including PV interneurons(Christophe et al. 2002; Letzkus et al. 2011; Arroyo et al.2012; Jiang et al. 2013; Lee et al. 2014). Consistent with this,we observed that foot-shocks cause a strong inhibitionof firing in PV interneurons in both anaesthetized andfreely behaving animals (Fig. 3Cc). In turn, this led toa marked disinhibition of pyramidal neurons duringand after the foot-shock (Fig. 3Cd). A second source ofdisinhibition may derive from direct cholinergic inhibitionof ongoing firing in layer 1 neurogliaform cells, whichhas recently been demonstrated in vitro (Brombas et al.2014). Thus, the population of foot-shock inhibited layer 1interneurons may correspond to neurogliaform cells,which provide strong direct inhibition to pyramidalneurons in layer 2/3 and layer 5 (Chu et al. 2003; Jianget al. 2013; Brombas et al. 2014; Lee et al. 2014). Inter-estingly, rapid recruitment by aversive stimuli has alsobeen observed for VIP interneurons in auditory cortex(Pi et al. 2013). Given that VIP interneurons expresscholinergic receptors (Porter et al. 1999), and can befired by acetylcholine in vivo (Alitto & Dan, 2013; Fuet al. 2014), it seems plausible that their activationduring aversive stimulation is mediated by acetylcholine,although this was not assessed directly. Taken together,these data demonstrate that one prominent network effect

of phasic acetylcholine release during aversive stimulationis disinhibition of pyramidal neurons via several parallelpathways.

Disinhibition of pyramidal neurons strongly booststheir responses to concomitantly presented auditorystimuli in both anaesthetized and freely behaving mice(Fig. 3F; cf. Froemke et al. 2007). This period ofenhanced firing is likely to induce synaptic plasticitythat may underlie the memory trace. A key pre-diction from this interpretation is that auditory cortexdisinhibition during the foot-shock is required forlearning at the behavioural level. To address this directly,we introduced channelrhodopsin-2 into PV interneurons,and used optogenetic activation of these neurons duringthe foot-shock to counteract the observed inhibition(Fig. 4A–C). This was performed in freely behaving miceduring fear conditioning to complex auditory stimuli,which have been proposed to require auditory cortexfor learning (LeDoux, 2000). When the memory of theseanimals was retrieved on the next day without optogeneticintervention, they displayed strongly reduced freezing inresponse to the conditioned stimulus (Fig. 4D), directlydemonstrating that auditory cortex disinhibition duringthe foot-shock is crucial for fear learning.

Summary and outlook

The studies reviewed here have advanced our under-standing of the mechanisms by which acetylcholinemodulates cortical circuit function during behaviour inseveral ways. First, they clearly demonstrate that inter-neurons are a prominent target of cholinergic modulation.The effects of acetylcholine, as predicted from slice work,strongly depend on interneuron type. In addition to the

B

ChR-2 expression in PV+−interneurons

Retrieval 24h without laser

Baseline CS− CS+

706050403020100

Free

zing

(%)

Sham Optogenetics Reconditioning

*** ***n = 8 each

C

DChR 2 expression in PV+

ACx

A

Laser 6 s

CS−

Foot-shock 1 sCS+

30 s...

...30 s

Figure 4. Auditory cortex disinhibition is requiredfor fear learningA, stimulation of ChR-2 expressing PV interneurons(green) in auditory cortex (red) via an optic fibre (blue). B,schematic illustration of optogenetic manipulation infreely behaving mice. C, differential fear conditioningprotocol using frequency-modulated sweeps asconditioned stimuli (CS) with optogenetic stimulationduring and for 5 s after every foot-shock. D, freezing in anovel context without laser stimulation 1 day afterconditioning. Compared to identically treated shaminjected littermates (black), virus-injected mice (blue)exhibit drastically reduced freezing to the CS+.Reconditioning without optogenetic stimulation yieldedstrongly enhanced freezing (red) that wasindistinguishable from sham. Grey background: CS–;green background: CS+. Adapted with permission fromLetzkus et al. 2011.



direct actions of acetylcholine, these in vivo studies havealso identified strong indirect effects, such as the inhibitionof PV and SOM interneurons by layer 1 and VIP inter-neurons (Letzkus et al. 2011; Alitto & Dan, 2013; Piet al. 2013; Fu et al. 2014). Second, in contrast to thetraditional view of neuromodulation as a relatively slowprocess, these results show that cholinergic signalling canrapidly fire certain interneuron types, thereby conveyingtime-locked information to neocortex (Letzkus et al. 2011;Arroyo et al. 2012; Alitto & Dan, 2013; Fu et al. 2014).Very similar data have recently been reported from thehippocampus, where aversive air puffs rapidly fire SOMinterneurons through acetylcholine release from medialseptum afferents (Lovett-Barron et al. 2014). Third, theinterneuron types displaying the strongest excitation byacetylcholine are layer 1 and VIP interneurons, which inturn preferentially contact other interneurons (Letzkuset al. 2011; Alitto & Dan, 2013; Pi et al. 2013; Fu et al.2014). This suggests that one important network effect ofphasic acetylcholine release is disinhibition of pyramidalneurons. A parallel disinhibitory mechanism that has beendescribed in vitro is mediated by activation of muscarinicreceptors on presynaptic PV interneuron terminals todirectly inhibit GABA release (Fukudome et al. 2004;Kruglikov & Rudy, 2008; Szabo et al. 2010), and aninteresting question for future experiments is under whichconditions this occurs in vivo. Disinhibition is an attractivemechanism to increase sensory responses during learningand active behavioural states since it is permissive for strongactivation of pyramidal neurons by sensory input, but doesnot cause firing in itself. In addition, disinhibition canincrease the gain of pyramidal neuron responses withoutaffecting stimulus selectivity (Fu et al. 2014).

In conclusion, differential recruitment of distinct inter-neuron types is emerging as an important mechanismthrough which acetylcholine can adjust the functionof neocortical circuits according to current behaviouralrequirements. Interestingly, several of the interneurontypes discussed here are also responsive to other neuro-modulators such as serotonin (Ferezou et al. 2002;Foehring et al. 2002; Lee et al. 2010). Important openquestions are therefore whether other neuromodulatorsare capable of mediating similarly rapid and specificrecruitment of interneurons (cf. Varga et al. 2009), andhow the effects of different modulators combine to shapethe function of neocortical circuits.

References

Alitto HJ & Dan Y (2013). Cell-type-specific modulation ofneocortical activity by basal forebrain input. Front SystNeurosci 6, 79.

Arroyo S, Bennett C, Aziz D, Brown SP & Hestrin S (2012).Prolonged disynaptic inhibition in the cortex mediated byslow, non-α7 nicotinic excitation of a specific subset ofcortical interneurons. J Neurosci 32, 3859–3864.

Bacci A, Huguenard JR & Prince DA (2005). Modulation ofneocortical interneurons: extrinsic influences and exercisesin self-control. Trends Neurosci 28, 602–610.

Bargmann CI (2012). Beyond the connectome: how neuro-modulators shape neural circuits. Bioessays 34, 458–465.

Bennett C, Arroyo S, Berns D & Hestrin S (2012). Mechanismsgenerating dual-component nicotinic EPSCs in corticalinterneurons. J Neurosci 32, 17287–17296.

Bennett C, Arroyo S & Hestrin S (2013). Subthresholdmechanisms underlying state-dependent modulation ofvisual responses. Neuron 80, 350–357.

Brombas A, Fletcher LN & Williams SR (2014).Activity-dependent modulation of layer 1 inhibitoryneocortical circuits by acetylcholine. J Neurosci 34,1932–1941.

Christophe E, Roebuck A, Staiger JF, Lavery DJ, Charpak S &Audinat E (2002). Two types of nicotinic receptors mediatean excitation of neocortical layer I interneurons.J Neurophysiol 88, 1318–1327.

Chu Z, Galarreta M & Hestrin S (2003). Synaptic interactionsof late-spiking neocortical neurons in layer 1. J Neurosci 23,96–102.

David C, Schleicher A, Zuschratter W & Staiger JF (2007). Theinnervation of parvalbumin-containing interneurons byVIP-immunopositive interneurons in the primarysomatosensory cortex of the adult rat. Eur J Neurosci 25,2329–2340.

Disney AA, Aoki C & Hawken MJ (2007). Gain modulation bynicotine in macaque v1. Neuron 56, 701–713.

Fanselow EE, Richardson KA & Connors BW (2008). Selective,state-dependent activation of somatostatin-expressinginhibitory interneurons in mouse neocortex. J Neurophysiol100, 2640–2652.

Ferezou I, Cauli B, Hill EL, Rossier J, Hamel E & Lambolez B(2002). 5-HT3 receptors mediate serotonergic fast synapticexcitation of neocortical vasoactive intestinal peptide/cholecystokinin interneurons. J Neurosci 22, 7389–7397.

Foehring RC, van Brederode JF, Kinney GA & Spain WJ (2002).Serotonergic modulation of supragranular neurons in ratsensorimotor cortex. J Neurosci 22, 8238–8250.

Fries P (2009). Neuronal gamma-band synchronization as afundamental process in cortical computation. Annu RevNeurosci 32, 209–224.

Froemke RC, Merzenich MM & Schreiner CE (2007). Asynaptic memory trace for cortical receptive field plasticity.Nature 450, 425–429.

Fu Y, Tucciarone JM, Espinosa JS, Sheng N, Darcy DP, NicollRA, Huang ZJ & Stryker MP (2014). A cortical circuit forgain control by behavioral state. Cell 156, 1139–1152.

Fukudome Y, Ohno-Shosaku T, Matsui M, Omori Y, FukayaM, Tsubokawa H, Taketo MM, Watanabe M, Manabe T &Kano M (2004). Two distinct classes of muscarinic action onhippocampal inhibitory synapses: M2-mediated directsuppression and M1/M3-mediated indirect suppressionthrough endocannabinoid signalling. Eur J Neurosci 19,2682–2692.

Gentet LJ, Kremer Y, Taniguchi H, Huang ZJ, Staiger JF &Petersen CC (2012). Unique functional properties ofsomatostatin-expressing GABAergic neurons in mousebarrel cortex. Nat Neurosci 15, 607–612.



Gil Z, Connors BW & Amitai Y (1997). Differential regulationof neocortical synapses by neuromodulators and activity.Neuron 19, 679–686.

Goard M & Dan Y (2009). Basal forebrain activation enhancescortical coding of natural scenes. Nat Neurosci 12,1444–1449.

Gulledge AT, Park SB, Kawaguchi Y & Stuart GJ (2007).Heterogeneity of phasic cholinergic signalling in neocorticalneurons. J Neurophysiol 97, 2215–2229.

Gunaydin LA, Yizhar O, Berndt A, Sohal VS, Deisseroth K &Hegemann P (2010). Ultrafast optogenetic control. NatNeurosci 13, 387–392.

Harris KD & Thiele A (2011). Cortical state and attention. NatRev Neurosci 12, 509–523.

Hasselmo ME (2006). The role of acetylcholine in learning andmemory. Curr Opin Neurobiol 16, 710–715.

Hasselmo ME & Giocomo LM (2006). Cholinergic modulationof cortical function. J Mol Neurosci 30, 133–135.

Herrero JL, Roberts MJ, Delicato LS, Gieselmann MA, Dayan P& Thiele A (2008). Acetylcholine contributes throughmuscarinic receptors to attentional modulation in V1.Nature 454, 1110–1114.

Isaacson JS & Scanziani M (2011). How inhibition shapescortical activity. Neuron 72, 231–243.

Jiang X, Wang G, Lee AJ, Stornetta RL & Zhu JJ (2013). Theorganization of two new cortical interneuronal circuits. NatNeurosci 16, 210–218.

Kawaguchi Y (1997). Selective cholinergic modulation ofcortical GABAergic cell subtypes. J Neurophysiol 78,1743–1747.

Kawai H, Lazar R & Metherate R (2007). Nicotinic control ofaxon excitability regulates thalamocortical transmission. NatNeurosci 10, 1168–1175.

Kepecs A & Fishell G (2014). Interneuron cell types are fit tofunction. Nature 505, 318–326.

Kerr JN & Denk W (2008). Imaging in vivo: watching the brainin action. Nat Rev Neurosci 9, 195–205.

Kimura F & Baughman RW (1997). Distinct muscarinicreceptor subtypes suppress excitatory and inhibitorysynaptic responses in cortical neurons. J Neurophysiol 77,709–716.

Kruglikov I & Rudy B (2008). Perisomatic GABA release andthalamocortical integration onto neocortical excitatory cellsare regulated by neuromodulators. Neuron 58, 911–924.

LeDoux JE (2000). Emotion circuits in the brain. Annu RevNeurosci 23, 155–184.

Lee AJ, Wang G, Jiang X, Johnson SM, Hoang ET, Lante F,Stornetta RL, Beenhakker MP, Shen Y & Zhu JJ (2014).Canonical organization of layer 1 neuron-led corticalinhibitory and disinhibitory interneuronal circuits. CerebCortex DOI: 10.1093/cercor/bhu020.

Lee S, Hjerling-Leffler J, Zagha E, Fishell G & Rudy B (2010).The largest group of superficial neocortical GABAergicinterneurons expresses ionotropic serotonin receptors.J Neurosci 30, 16796–16808.

Lee S, Kruglikov I, Huang ZJ, Fishell G & Rudy B (2013). Adisinhibitory circuit mediates motor integration in thesomatosensory cortex. Nat Neurosci 16, 1662–1670.

Lee SH & Dan Y (2012). Neuromodulation of brain states.Neuron 76, 209–222.

Letzkus JJ, Wolff SB, Meyer EM, Tovote P, Courtin J, Herry C &Luthi A (2011). A disinhibitory microcircuit for associativefear learning in the auditory cortex. Nature 480,331–335.

Lovett-Barron M, Kaifosh P, Kheirbek MA, Danielson N,Zaremba JD, Reardon TR, Turi GF, Hen R, Zemelman BV &Losonczy A (2014). Dendritic inhibition in the hippocampussupports fear learning. Science 343, 857–863.

McCormick DA & Prince DA (1985). Two types of muscarinicresponse to acetylcholine in mammalian cortical neurons.Proc Natl Acad Sci U S A 82, 6344–6348.

Miranda MI, Ramirez-Lugo L & Bermudez-Rattoni F (2000).Cortical cholinergic activity is related to the novelty of thestimulus. Brain Res 882, 230–235.

Niell CM & Stryker MP (2010). Modulation of visual responsesby behavioral state in mouse visual cortex. Neuron 65,472–479.

Paolone G, Lee TM & Sarter M (2012). Time to pay attention:attentional performance time-stamped prefrontalcholinergic activation, diurnality, and performance.J Neurosci 32, 12115–12128.

Parikh V, Kozak R, Martinez V & Sarter M (2007). Prefrontalacetylcholine release controls cue detection on multipletimescales. Neuron 56, 141–154.

Pfeffer CK, Xue M, He M, Huang ZJ & Scanziani M (2013).Inhibition of inhibition in visual cortex: the logic ofconnections between molecularly distinct interneurons. NatNeurosci 16, 1068–1076.

Pi HJ, Hangya B, Kvitsiani D, Sanders JI, Huang ZJ & Kepecs A(2013). Cortical interneurons that specialize in disinhibitorycontrol. Nature 503, 521–524.

Pinto L, Goard MJ, Estandian D, Xu M, Kwan AC, Lee SH,Harrison TC, Feng G & Dan Y (2013). Fast modulation ofvisual perception by basal forebrain cholinergic neurons. NatNeurosci 16, 1857–1863.

Poorthuis RB, Bloem B, Schak B, Wester J, de Kock CP &Mansvelder HD (2013). Layer-specific modulation of theprefrontal cortex by nicotinic acetylcholine receptors. CerebCortex 23, 148–161.

Porter JT, Cauli B, Tsuzuki K, Lambolez B, Rossier J & AudinatE (1999). Selective excitation of subtypes of neocorticalinterneurons by nicotinic receptors. J Neurosci 19,5228–5235.

Reynolds JH & Chelazzi L (2004). Attentional modulation ofvisual processing. Annu Rev Neurosci 27, 611–647.

Richardson RT & DeLong MR (1990). Context-dependentresponses of primate nucleus basalis neurons in a go/no-gotask. J Neurosci 10, 2528–2540.

Sarter M & Bruno JP (1997). Cognitive functions of corticalacetylcholine: toward a unifying hypothesis. Brain Res BrainRes Rev 23, 28–46.

Sarter M, Parikh V & Howe WM (2009). Phasic acetylcholinerelease and the volume transmission hypothesis: time tomove on. Nat Rev Neurosci 10, 383–390.

Szabo GG, Holderith N, Gulyas AI, Freund TF & Hajos N(2010). Distinct synaptic properties of perisomaticinhibitory cell types and their different modulation bycholinergic receptor activation in the CA3 region of themouse hippocampus. Eur J Neurosci 31,2234–2246.



Taniguchi H, He M, Wu P, Kim S, Paik R, Sugino K, KvitsianiD, Fu Y, Lu J, Lin Y, Miyoshi G, Shima Y, Fishell G, NelsonSB & Huang ZJ (2011). A resource of Cre driver lines forgenetic targeting of GABAergic neurons in cerebral cortex.Neuron 71, 995–1013.

Varga V, Losonczy A, Zemelman BV, Borhegyi Z, Nyiri G,Domonkos A, Hangya B, Holderith N, Magee JC & FreundTF (2009). Fast synaptic subcortical control of hippocampalcircuits. Science 326, 449–453.

Weinberger NM (2007). Auditory associative memory andrepresentational plasticity in the primary auditory cortex.Hear Res 229, 54–68.

Wilson FA & Rolls ET (1990a). Neuronal responses related toreinforcement in the primate basal forebrain. Brain Res 509,213–231.

Wilson FA & Rolls ET (1990b). Neuronal responses related tothe novelty and familarity of visual stimuli in the substantiainnominata, diagonal band of Broca and periventricularregion of the primate basal forebrain. Exp Brain Res 80,104–120.

Woolf NJ (1991). Cholinergic systems in mammalian brain andspinal cord. Prog Neurobiol 37, 475–524.

Xiang Z, Huguenard JR & Prince DA (1998). Cholinergicswitching within neocortical inhibitory networks. Science281, 985–988.

Zhang F, Aravanis AM, Adamantidis A, de Lecea L & DeisserothK (2007). Circuit-breakers: optical technologies for probingneural signals and systems. Nat Rev Neurosci 8, 577–581.

Additional information

Competing interests

None declared.

Funding

Our research is supported by the Max Planck Society,the European Research Council, the Swiss National ScienceFoundation, the Netherlands Organization for ScientificResearch (NWO Rubicon, 825.13.015) and the BoehringerIngelheim Fonds.


Cholinergic circuit modulation through differential...

Documents

Transcript of Cholinergic circuit modulation through differential...