Modality-specific thalamocortical inputs instruct the identity of postsynaptic L4 neurons

LETTERdoi:10.1038/nature13390

Modality-specific thalamocortical inputs instructthe identity of postsynaptic L4 neuronsGabrielle Pouchelon1, Frederic Gambino1{, Camilla Bellone1, Ludovic Telley1, Ilaria Vitali1, Christian Luscher1,2,3,Anthony Holtmaat1 & Denis Jabaudon1,2,3

During development, thalamocortical (TC) input has a critical rolein the spatial delineation and patterning of cortical areas1–6, yet theunderlying cellular and molecular mechanisms that drive corticalneuron differentiation are poorly understood. In the primary (S1)and secondary (S2) somatosensory cortex, layer 4 (L4) neurons receivemutually exclusive input originating from two thalamic nuclei7,8: theventrobasalis (VB), which conveys tactile input9,10, and the posteriornucleus (Po), which conveys modulatory and nociceptive input11–14.Recently, we have shown that L4 neuron identity is not fully com-mitted postnatally15, implying a capacity for TC input to influencedifferentiation during cortical circuit assembly. Here we investigatewhether the cell-type-specific molecular and functional identity ofL4 neurons is instructed by the origin of their TC input. Genetic abla-tion of the VB at birth resulted in an anatomical and functional rewir-ing of Po projections onto L4 neurons in S1. This induced acquisitionof Po input led to a respecification of postsynaptic L4 neurons, whichdeveloped functional molecular features of Po-target neurons whilerepressing VB-target traits. Respecified L4 neurons were able to res-pond both to touch and to noxious stimuli, in sharp contrast to thenormal segregation of these sensory modalities in distinct corticalcircuits. These findings reveal a behaviourally relevant TC-input-type-specific control over the molecular and functional differenti-ation of postsynaptic L4 neurons and cognate intracortical circuits,which instructs the development of modality-specific neuronal andcircuit properties during corticogenesis.

Within S1, VB axons target L4 neurons, forming cortical barrels, andPo axons target L5A and L1 neurons; in S2, however, Po axons target L4neurons (Fig. 1a, b)7,8,16,17. Molecular distinctions between L4 neuronsin S1 and S2, such as Rorb expression18 (Fig. 1c), may therefore in part bedriven by their distinct TC inputs. To investigate TC-type-specific con-trols over L4 neuron identity, we genetically ablated the VB at birthby generating transgenic Slc6a4::Cre/Rosa26::stopfloxDTA mice (vb2

mice)19,20, leading to death of VB neurons between postnatal day (P)0and P4 and lack of formation of associated S1 whisker-mapped corticalbarrels (Fig. 1d–h and Extended Data Figs 1, 2a–c). Remarkably, S1L4remained richly innervated by TC terminals despite the absence of VBaxons, as shown using the pan-TC presynaptic marker VGLUT2 (ref. 18)(Fig. 1i), without evidence for secondary cell death (Extended DataFig. 2d–g). These terminals did not belong to residual VB axons, as theywere still present at P23 (Extended Data Fig. 2b) and did not express theVB-specific markers 5HTT (5-hydroxytryptamine transporter) and GSBS(G-substrate)21 (Fig. 1j, k). These data indicate that S1L4 neurons stillreceive TC input in the absence of VB.

To identify the origin of these ectopic TC projections, we retrogradelylabelled TC neurons from S1, revealing that Po is the exclusive source ofTC input to S1 in vb2 mice (Extended Data Fig. 3a–f’). Povb2 neuronswere undistinguishable from control Po neurons by microarray compar-ative gene expression analysis (Extended Data Fig. 3g), demonstratingthat bona fide Po neurons are the source of S1L4 TC input in vb2 mice.To directly visualize aberrant Povb2 projections, we next anterogradely

1Department of Basic Neurosciences, Faculty of Medicine, University of Geneva, CH-1211 Geneva, Switzerland. 2Clinic of Neurology, Department of Clinical Neurosciences, Geneva University Hospital, CH-1211 Geneva, Switzerland. 3Institute of Genetics & Genomics in Geneva (iGE3), University of Geneva, CH-1211 Geneva, Switzerland. {Present address: Interdisciplinary Institute for NeuroScience, CNRSUMR 5297, 33077 Bordeaux, France.

Po

VB

L2/3

L4

L5

Controle

Po Po

Cyt

oxid

ase

Cyt

oxid

ase

Po

dLG

GS

BS

VB

Po

Slc6a4Cre tdTomato

dS1 S2S1 S2

L2/3

L4

Po inputVB input

L5AL5B

L6

h

VBPo

dLG dLG

Po

Control

CA

LB

2

S1 S2S1 S2

ba

L2/3

L4L5A

L5B

L6

L4

L6

L2/3

vb–

vb– vb–

vb– vb–

ant

lat

i

GS

BS

L4

L6

L2/3

5H

TT

Control

c

gf

j k

L2/3

L4L5A

L5B

L6

L5A

L5B

L5A

L5B

VG

LU

T2

Slc6a4 Cre

loxPRosa26

loxPSTOPX DTA

Thalamus S1

dLG

Po

VB

h Control Control vb–Control vb–Control

P7

L4

L4 S1

Po

VB

S2

P0 P7

P4P7 P7

P7

P77P

Rorb Rorb

P10200 μm

400 μm

5HTT

Ant

labelling

500 μm 200 μm

Figure 1 | TC input to S1L4 is preserved in vb2 mice. a, b, VB and Po TCaxons have specific projections to S1 and S2. c, Rorb expression is area-specificin L4 neurons (from Allen Brain Atlas). d, tdTomato reporter expressiondriven by Slc6a4Cre shows VB-specific recombination. e–h, vb2 mice lack VB

(e–g) and lack presynaptic barrels in S1 (h). Ant, anterograde; Cyt oxidase,cytochrome oxidase; dLG, dorsolateral geniculate nucleus. i, Presynaptic TCterminals are still present in S1vb2 cortex using the pan-TC presynaptic markerVGLUT2. j, k, VB-specific presynaptic markers are lacking in S1vb2.

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 1

Macmillan Publishers Limited. All rights reserved©2014

www.nature.com/doifinder/10.1038/nature13390

labelled Po and VB axons in control mice and Po axons in vb2 mice.Control VB axons branched within S1L4 and Po axons were excludedfrom this layer, whereas conversely, in S2, where VB axons were not found,Po axons branched within L4 (Fig. 2a, b, d). By contrast, in S1vb2, Poneurons projected densely to L4 (Fig. 2c, d; P , 0.05 for Po control (PoCtrl)versus Povb2 axonal length in L4, Student’s t-test, n 5 3 control and n 5 4vb2 labellings). Po axons formed functional synapses onto S1L4vb2 neu-rons, as demonstrated by short-latency synaptic responses to optoge-netic stimulation of these axons, which were absent in control S1L4 barrelneurons (Fig. 2e, f). By contrast, Po-S2L4 connectivity was unchangedin vb2 cortex (Extended Data Fig. 3h, i), and Po rewiring in S1 was notobserved following focal VB ablation at P10 (Supplementary Note 1).Taken together, these results indicate that Po axons substitute for VBaxons in vb2 mice, providing a new source of presynaptic input to L4neurons.

We next investigated whether this VBRPo switch in TC input instructscell-type-specific developmental gene expression programs in postsyn-aptic L4 neurons. We first characterized the molecular identity of wild-type cortical neurons in L2/3, L4, and L5/6 of S1 and S2 using microarrayanalysis of microdissected samples22 at P10 (Fig. 3a, b and Extended DataFig. 4a, b). The genetic relatedness between these samples was deter-mined using unbiased cluster analysis, revealing three classes of tran-scriptional programs: L2/3-like, L4-like and L5/6-like. Analysis of S1L4vb2

samples revealed an L4-like transcriptional program, demonstrating thatTC input origin does not determine the laminar molecular identity ofL4 neurons (Fig. 3c).

We next investigated whether TC input origin instructs distinct dif-ferentiation programs in VB-receiving (S1L4) and Po-receiving (S2L4)

neurons. To examine whether the VBRPo switch in TC input repressedVB-target and induced Po-target differentiation programs, we first definedS1L4- and S2L4-type transcripts by identifying wild-type L4-specific geneswhose expression was mutually exclusive between S1 and S2 (Fig. 3d).Analysis of the expression of these two gene sets by S1L4vb2 neuronsrevealed a strong repression of S1L4-type genes and a correspondinginduction of S2L4-type genes (Fig. 3e–g; P , 0.0001 for S1L4 (n 5 74genes) and ,0.005 for S2L4 (n 5 74 genes) percentage expression level,paired Student’s t-test). This respecification of molecular identity wasfurther examined using in situ hybridization for select gene candidatesincluding the S1L4-type transcript Rorb, which was strongly decreasedin S1L4vb2 neurons, and the S2L4-type transcript Cdh8, which was upreg-ulated (Fig. 3h, i and Extended Data Fig. 4c–e). VB input-dependentchanges in S1L4 gene expression coincided with the postnatal matura-tion of VB-L4 synapses, and failed to occur in S1L4vb2 neurons, result-ing in an S2L4-type developmental gene expression program (ExtendedData Figs 1a–i, 5 and Supplementary Note 2).

We next investigated the respective contribution of VB input loss andPo input acquisition to these genetic changes. Comparison of vb2 mice

L4

L2/3

L5

L6VB

co

ntr

ol

Po

co

ntr

ol

a

b

c

L4L2/3

L5

L6

L1

L1

S1 VB

S1 Po

S1 Povb–

S2 VB

L4L2/3

L5

L6

L4L2/3

L5

L6

L1

L1

L4L2/3

L5

L6

L1

From VB

from VP

From Po From Po

Po

vb–

Bin

num

ber

S1PoS1VB S1Povb–

1

2

3

4

5

Bin

n

S2 Po

d

P7

From Povb– CUX1

S2Po

**

400 20 % axonal length in bin

Po

1 mm

e

ChR

2-G

FP

Co

nnecte

d n

euro

ns

(ChR

2-i

nd

uced

EP

SC

s) (%

)

200 ms

100

80

60

40

20

0

L4 L4vb–

n =

5

n =

10

Ctr

l L4

L4

vb–

Ctr

l L

5A

Po photostim.

Po

VB

20

pA

US

co

ronal

P0

Cx

DG P20

Po

L1

L2/3

L5AL4

L6

Po

L5B

Control vb–

VB

f

**

1

2

3

4

5

1

2

3

4

5

12345

1

2

3

4

5

1 mm 100 μm

100 μm

Figure 2 | Po TC axons are rewired onto S1L4 neurons in vb2 mice.a, b, Anterograde labelling in control mice showing arborization of VB axonswithin S1L4 (a, left) and exclusion of Po axons from this layer (b, left). In S2,VB axons are absent (a, right) whereas Po axons arborize in S2L4 (b, right).c, Povb2 axons arborize within S1L4. d, Quantification. **P , 0.005, Student’st-test, n 5 3 control (Ctrl) and n 5 4 vb2 injections. e, Ultrasound (US)-guidedmicroinjection of a AAV-ChR2-GFP virus into Po. Red arrowheads indicatelocation of the micropipette shaft. f, Optogenetic stimulation of Po axons elicitsexcitatory postsynaptic currents (EPSCs) in S1L5A neurons but not S1L4neurons, whereas EPSCs are evoked in S1L4vb2 neurons. **P , 0.005,Student’s t-test.

a c

S1 S2

7426

L4- specific

S1L4

S2L4

S1L4

vb–

S1L4 genes S2L4 genes

S1L4

vb–

exp

ressio

n

S1L4 e

xp

ressio

n level (%

)

*** **

S1L4 expression

e f

L2/3

L5/6

L4

S1 S2 S1vb–

From VB From Po From Po

1.63–1.63 0

L4- specific

100

150

0

50

2040

n g

enes 20

40

20

40

0

S1L4 expression

100

10

20

200

20

40

74

S1

L4

S2

L4

30

10

10 30 5 15

5

15

g

b

L2/3

L4

L5

S1

0% +50%–50%

Relative expression level

L4vb–

*** *****

d

0

PC1 (48%)

PC

2 (2

9%

)

0

S2L5/6

S1L5/6

S2L2/3

S1L2/3

S2L4

S1L4

S1L4vb–

–9 9

**

*

P10

S1L4 genes S2L4 genes

*

j

Sam

ple

s

GenesL2/3 L4 L5

6

–6

12

–12

18–18

Expression level

0

S1

L4

S2

L4

Ror

b

Control vb– vb– vb–

S1

S2

h

P23

Cd

h8

P7

Control

S1

S2

Ctrl S1 Ctrl S2 Ctrl S1 Ctrl S2

L4

L2/3

L5

L4

L2/3

L5

S1vb– S1vb–

S1vb– S1vb–

i

Exp

ressio

n

GFP VGLUT2

Ctr

lN

pas4

*

Control Npas4GFP

Ctrl

Dend

rite

s in h

ollo

w (%

)

0

100

Ctr

l

Np

as4

epor E14.59

5

0 Npas4

P7

GFP

GFPVGLUT2

L4

Septa

Ho

llow

1 mm

200 μm

200 μm

20 μm

decre

ased

incre

ased

Figure 3 | VBRPo switch in input congruently respecifies S1L4vb2

neurons. a, Schematic experimental design. b, Validation of sample collectionwith known marker genes for L2/3, L4 and L5/6 neurons (see ref. 22 andMethods). c, Principal component (PC) analysis plot of sample-specific geneexpression showing clustering of L2/3 (triangles), L4 (circles) and L5/6(squares) transcriptional programs. d, Delineation of S1L4-type and S2L4-typetranscripts. e, Heatmap representation showing repression of a subset ofS1L4-type genes (blue asterisk) and induction of S2L4-type genes (red asterisk)in L4vb2 neurons. f, Levels of expression of S1L4-type genes (green: full,decreased; open, increased) and S2L4-type genes (purple: full, increased; open,decreased) in L4vb2 neurons. **P , 1024, Student’s t-test. g, 62 out of 74S1L4-type genes are downregulated in S1L4vb2 neurons; 50 out of 74 S2L4-typegenes are upregulated in S1L4vb2 neurons. ***P , 1025, **P , 0.005, pairedStudent’s t-test; n 5 74 S1L4-type and 74 S2L4-type genes. h, i, In situhybridization illustrating cell-type-specific downregulation of Rorb (h), andupregulation of Cdh8 (i). j, Overexpression of an Npas4GFP-expressing constructinto S1L4 neurons disrupts the polarization of dendrites towards barrel hollows(that is, VB axon-containing barrel centres) at P7. *P , 0.05, n 5 19 Ctrl andn 5 10 Npas41 neurons, ANOVA. Error bars denote mean 6 s.e.m.

RESEARCH LETTER

2 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4


with Dlx5/6::Cre/Celsr3flox mice, which entirely lack TC input23, revealedthat each input type instructs the differentiation program of S1L4vb2

neurons, as some genes were strictly VB-input-dependent, whereas otherswere Po-input-dependent (Extended Data Fig. 6 and SupplementaryNote 3). Together, these data indicate that VB and Po axons exert input-type-specific controls over S1L4 neuron genetic identity.

We next examined the functional role of the top input-dependent S1L4transcripts, which together formed a core set of genes involved in neuritedifferentiation and polarization (Extended Data Table 1). The transcriptionfactors Npas4 (ref. 24) and Zbtb20 (ref. 25) and the extracellular pro-tein Sema3a (ref. 26) were functionally characterized using in vitro andin vivo morphological analysis of embryonic day (E)14.5-electroporatedL4 neurons. Each of these transcripts controlled neurite polarizationin vitro and postnatal orientation of S1L4 neuron primary dendritestowards VB axons in vivo, a key requisite for columnar assembly of cor-tical circuits (Fig. 3j, Extended Data Fig. 7 and Supplementary Note 4).Thus, VB input exerts a genetic control over the development of criticalmorphological circuit-related properties of S1L4 neurons.

To further investigate the role of VB input in intracortical circuit assem-bly, we determined excitatory/inhibitory (E/I) feed-forward responsesof L4 neurons to optogenetic stimulation of VB (for S1L4 neurons) andPo axons (for S2L4 and S1L4vb2, Extended Data Fig. 8a). Supporting aTC-type-specific control over intracortical circuit assembly, S1L4vb2

E/I responses were increased to S2L4 levels (Extended Data Fig. 8b–d;P , 0.05 for S1L4 versus S2L4 and S1L4 versus S1L4vb2; P 5 0.4 for S2L4versus S1L4vb2, analysis of variance (ANOVA), n 5 5 Ctrl, 7 S2L4 and6 S1L4vb2 neurons). VBRPo switch in input thus instructs correspond-ing downstream changes in intracortical circuit properties.

As VB neurons normally transmit tactile information9,10 and Po neu-rons nociceptive information12–14, we examined whether the VBRPoswitch in input led to a congruent shift in the functional identity of S1L4neurons (Fig. 4a). The trigeminal principalis nucleus (PrV), which pro-vides input to VB, was notably atrophied in vb2 mice, whereas Po afferentpathways remained unchanged, suggesting that the locus of the plas-ticity in vb2 mice is essentially thalamocortical (Extended Data Fig. 9and Supplementary Note 5). In control mice, single-whisker environ-mental exploration led to activation of S1L4 neurons in the corresponding

barrel, as reported by expression of the immediate-early gene c-fos (alsoknown as Fos)14,27 (Fig. 4b). S1L4vb2 neurons were also activated by thistask, although more diffusely, consistent with the broader distributionof presynaptic Povb2 terminals and S2-like increases in S1L4vb2 E/Iresponses (Supplementary Note 6). Back-and-forth deflections of singlewhiskers in head-fixed mice evoked focal intrinsic optical signal res-ponses in S1 (Fig. 4c)28, which were weaker and more variable than incontrol mice, suggesting asynchronous cortical activation. Accordingly,in vivo intracellular recordings revealed a ,3-fold increase in the jitterof L2/3-evoked responses to whisker deflections and prolonged onsetlatencies in vb2 mice (Fig. 4d, P , 0.05 for jitter and latency, Mann–Whitney rank-sum test, n 5 7 Ctrl and 6 vb2 cells). Consistent withthis degradation in the coding of input signals, vb2 mice were distinctlyimpaired at tasks requiring haptic skills and fine sensorimotor coordi-nation (Fig. 4e and Supplementary Note 7). Together, these findingsreflect the low fidelity of spatial and temporal encoding in Po ascendingpathways29 and indicate that Po-input-receiving S1L4vb2 neurons stillrespond, albeit less reliably, to tactile stimuli in vivo.

To investigate whether respecified S1L4vb2 neurons acquired theability to respond to noxious stimuli, we performed a microinjection ofcapsaicin, an algogenic chemical, into the whisker pad of vb2 mice14. Incontrol mice, S1L5A neurons and S2L4 neurons were strongly activated,as revealed by c-Fos expression, while S1L4 neurons remained inactive,reflecting the normal target specificity of Po projections14. By contrast,S1L4vb2 neurons were activated by noxious stimulation, in a patternsimilar to that of S2L4 of control mice (Fig. 4f; P , 0.001 for L4 versusL4vb2, Student’s t-test; n 5 3 Ctrl and n 5 3 vb2 mice). Consistent withan expanded L4 population response to Po input, vb2 mice displayedincreased behavioural signs of pain following capsaicin injection andshortened flick latencies to focal thermic tail stimulation (Supplemen-tary Videos 1 and 2, Fig. 4g and Supplementary Note 7). Taken together,these data indicate that rewiring of Po input onto VB targets results inan abnormal convergence of normally segregated tactile and nocicep-tive sensory modalities onto S1L4 neurons.

Our findings reveal that distinct TC inputs exert modality-specificcontrols over the molecular identity and function of postsynaptic L4neurons. VB and Po inputs only affected a specific subset of S1L4- and

S1 control

L2/3

L5A

L4

L6

L5B

S2 control f

No

cic

ep

tio

n (P

o)

c-Fos

100

c-F

os c

ells

(% to

tal)

80

60

0

L5A

L4

L5A

L4

S1 Ctrl

40

20

***

**

c

CAM

Single-whisker stim.

1 sec 8 Hz

S1 optical imaging in vivo

Co

ntr

ol

R

LM

C

S1 controlb

Exp

lora

tio

n (V

B) L2/3

L5

L4

L6

c-Fos

S1vb– S1vb–

S1vb–

Activated whisker Control sideActivated whisker

10 m

V

10 ms

Co

ntr

ol

d S1 whole cell in vivo

vb–

vb–

*vb–

*15

10

5

0

20

Dela

y (m

s)

Ctrl

Ctrl

Jitte

r (m

s)

4

2

1

0

5

3

S1

S1vb–

Co

ntr

ol

vb–

vb–

vb–

a

S2

S2

40

30

20

Num

ber

of

foo

t-fa

ults

10

Ctrl vb– Ctrl vb– Ctrl vb– Ctrl vb–0

Grid walk

*

e

13 4

To

tal cro

sses (%

)

100

80

60

40

20

0

Fail

*Gap cross

Rem

oval s

ucess ra

te (%

)

40

30

0

20

First

co

nta

ct

late

ncy (s)

10

Hindpaw patch removal

**

12 5 12 5

g Tail-flick latency

To

tal events

(%

)

100

80

60

40

20

0

Flick latency (s)

0.5 1 2 3 4 5

vb–

Ctrl

**

1 2 3 4 s

20%

10%

100 ms

7 6

100

80

60

40

20

0

44

5

100 μm

100 μm

250 μm

***

Figure 4 | Functional convergence of normally segregated sensorymodalities in vb2 mice. a, S1L4vb2 neurons acquire S2L4-type functionalproperties. b, S1L4vb2 neurons are activated during environmental exploration.c, Single-whisker stimulation elicits weak focal S1 responses in vb2 mice.d, EPSCs from sample L2/3 neurons. Delay and increased jitter of responses towhisker deflections in vb2 mice (*P , 0.05, Mann–Whitney rank-sum test,n 5 7 Ctrl and 6 vb2 neurons). e, Haptic skills and fine sensorimotorcoordination are impaired in vb2 mice (see Supplementary Note 7). *P , 0.05,

**P 5 0.008, ***P , 0.0005, Student’s t-test. f, Noxious stimulation of thewhisker-pad activates S1L4vb2 neurons but not S1L4 neurons (n 5 3 Ctrl andn 5 3 vb2 mice;*P , 0.05, **P , 0.005, Student’s t-test) (L4 versus L5A in vb2

mice, P , 0.01; L4 versus L5A in control mice, P , 0.005; L4 versus L4vb2,P , 0.005). g, vb2 mice have shorter flick latencies upon focal thermicstimulation of the tail. ** P , 0.005, Kolmogorov–Smirnov test. Error barsdenote mean 6 s.e.m.

LETTER RESEARCH

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 3


S2L4-type genes, such that the identity of S1L4vb2 neurons was inter-mediate between these two neuronal types. This intermediate geneticidentity is reminiscent of the cytoarchitectonically ‘hybrid’ visual cortexfound following in utero bilateral enucleation in monkeys30, suggestingthat input-type-dependent specification may have contributed to cor-tical neuron diversity during evolution.

Po-dependent ‘non-lemniscal’ somatosensory pathways are thoughtto be evolutionary older than VB-dependent ‘lemniscal’ pathways (seeref. 14 and references therein). Emergence of a specialized VB-dependentpathway during evolution suggests that VB axons were at competitiveadvantage over pre-existing Po axons. Supporting this possibility, ourfindings reveal that Po axons invade VB targets in vb2 cortex, and VBand Po axons co-mingle in the early postnatal cortex before segregatinginto their anatomically complementary patterns16. Therefore, evolu-tionary and developmental segregation of tactile and nociceptive path-ways may have occurred via hierarchical interactions between VB andPo inputs.

Finally, the circuit-specific transcriptional controls described hereprovide a powerful mechanism matching terminal neuronal differenti-ation to specific functional constraints. Whereas cell-intrinsic differen-tiation programs initially define neuronal permissiveness to distinctinputs15, reciprocally, these inputs differentially instruct gene expressionprograms in their targets. This crosstalk between gene expression andcircuit connectivity may therefore act to orchestrate the assembly of cog-nate neurons into functionally specialized pathways during development.

METHODS SUMMARYAnimals. Compound mice were generated as described in Methods.Histology. Brains were fixed and stained using standard methods14,15.Anterograde and retrograde labelling. Retrograde cortical labelling was performedusing stereotaxic FluoroGold microinjections. Anterograde cortical labelling wasperformed on fixed brains using NeuroVue Dye Filter Red (MTTI, FS-1002) insertedinto the VB or the Po.Microdissection and microarray analysis. L2/3, L4 and L5/6 in S1 and S2 of P10 con-trols and vb2 mice were microdissected. After amplification and labelling of extractedRNA, hybridization onto Affymetrix Mouse Gene 1.0 ST arrays was performed.ChR2 injections and electrophysiology. A ChR2-Venus-expressing adenovirus(AAV5-hSyn-hChR2 H134R) was used. Electrophysiology recordings were per-formed at 3 weeks of age15.In utero electroporations. These were performed as described in ref. 15.Tactile and noxious stimulations. These were performed as detailed in ref. 14.In vivo electrophysiology. Intrinsic imaging and whole-cell recordings were per-formed as detailed in ref. 28.Behavioural tests. These were performed as described in the Methods.

Online Content Any additional Methods, Extended Data display items and SourceData are available in the online version of the paper; references unique to thesesections appear only in the online paper.

Received 11 October 2013; accepted 22 April 2014.

Published online 14 May 2014.

1. Sur, M., Garraghty, P. E. & Roe, A. W. Experimentally induced visual projections intoauditory thalamus and cortex. Science 242, 1437–1441 (1988).

2. Dehay, C., Horsburgh, G., Berland, M., Killackey, H. & Kennedy, H. Maturation andconnectivity of the visual cortex in monkey is altered by prenatal removal of retinalinput. Nature 337, 265–267 (1989).

3. Katz, L. C. & Shatz, C. J. Synaptic activity and the construction of cortical circuits.Science 274, 1133–1138 (1996).

4. Miyashita-Lin, E. M., Hevner, R. F., Wassarman, K. M., Martinez, S. & Rubenstein,J. L. R. Early neocortical regionalization in the absence of thalamic innervation.Science 285, 906–909 (1999).

5. Chou, S.-J. et al.Geniculocortical input drivesgenetic distinctionsbetweenprimaryand higher-order visual areas. Science 340, 1239–1242 (2013).

6. Li, H. et al. Laminar and columnar development of barrel cortex relies onthalamocortical neurotransmission. Neuron 79, 970–986 (2013).

7. Wimmer, V.C., Bruno,R.M., deKock,C.P. J., Kuner, T.&Sakmann, B.DimensionsofaprojectioncolumnandarchitectureofVPMandPOm axons in rat vibrissal cortex.Cereb. Cortex 20, 2265–2276 (2010).

8. Ohno, S. et al. A morphological analysis of thalamocortical axon fibers of ratposterior thalamic nuclei: a single neuron tracing study with viral vectors. Cereb.Cortex 22, 2840–2857 (2012).

9. Nicolelis, M. A. L. Computing with thalamocortical ensembles during differentbehavioural states. J. Physiol. (Lond.) 566, 37–47 (2005).

10. Yu, C., Derdikman, D., Haidarliu, S. & Ahissar, E. Parallel thalamic pathways forwhisking and touch signals in the rat. PLoS Biol. 4, e124 (2006).

11. Ahissar, E., Sosnik, R. & Haidarliu, S. Transformation from temporal to rate codingin a somatosensory thalamocortical pathway. Nature 406, 302–306 (2000).

12. Gauriau, C. & Bernard, J.-F. Posterior triangular thalamic neurons conveynociceptive messages to the secondary somatosensory and insular cortices in therat. J. Neurosci. 24, 752–761 (2004).

13. Masri, R. et al. Zona incerta: a role in central pain. J. Neurophysiol. 102, 181–191(2009).

14. Frangeul, L. et al. Specific activation of the paralemniscal pathway duringnociception. Eur. J. Neurosci. http://dx.doi.org/10.1111/ejn.12524 (2 March2014).

15. De la Rossa, A. et al. In vivo reprogramming of circuit connectivity in postmitoticneocortical neurons. Nature Neurosci. 16, 193–200 (2013).

16. Kichula, E. A.& Huntley,G.W.Developmental andcomparative aspects ofposteriormedial thalamocortical innervation of the barrel cortex in mice and rats. J. Comp.Neurol. 509, 239–258 (2008).

17. Viaene, A. N., Petrof, I. & Sherman, S. M. Properties of the thalamic projection fromthe posterior medial nucleus to primary and secondary somatosensory cortices inthe mouse. Proc. Natl Acad. Sci. USA 108, 18156–18161 (2011).

18. Jabaudon, D., Shnider, S. J., Tischfield, D. J., Galazo, M. J. & Macklis, J. D. RORbinduces barrel-like neuronal clusters in the developing neocortex. Cereb. Cortex22, 996–1006 (2012).

19. Narboux-Neme, N., Pavone, L. M., Avallone, L., Zhuang, X. & Gaspar, P. Serotonintransporter transgenic (SERTcre) mouse line reveals developmental targets ofserotonin specific reuptake inhibitors (SSRIs). Neuropharmacology 55, 994–1005(2008).

20. Brockschnieder, D. et al. Cell depletion due to diphtheria toxin fragment A afterCre-mediated recombination. Mol. Cell. Biol. 24, 7636–7642 (2004).

21. Endo, S. G-substrate: the cerebellum and beyond. Prog. Mol. Biol. Transl. Sci. 106,381–416 (2012).

22. Belgard, T. G. et al. A transcriptomic atlas of mouse neocortical layers. Neuron 71,605–616 (2011).

23. Zhou, L. et al. Early forebrain wiring: genetic dissection using conditional Celsr3mutant mice. Science 320, 946–949 (2008).

24. Yun, J. et al. Neuronal Per Arnt Sim (PAS) domain protein 4 (NPAS4) regulatesneurite outgrowth and phosphorylation of synapsin I. J. Biol. Chem. 288,2655–2664 (2013).

25. Nielsen, J. V., Blom, J. B., Noraberg, J. & Jensen, N. A. Zbtb20-induced CA1pyramidal neuron development and area enlargement in the cerebral midlinecortex of mice. Cereb. Cortex 20, 1904–1914 (2010).

26. Shelly, M. et al. Semaphorin3A regulates neuronal polarization by suppressingaxon formation and promoting dendrite growth. Neuron 71, 433–446 (2011).

27. Wagener, R. J., David, C., Zhao, S., Haas, C. A. & Staiger, J. F. The somatosensorycortex of reeler mutant mice shows absent layering but intact formation andbehavioral activation of columnar somatotopic maps. J. Neurosci. 30,15700–15709 (2010).

28. Gambino, F. & Holtmaat, A. Spike-timing-dependent potentiation of sensorysurround in the somatosensory cortex is facilitated by deprivation-mediateddisinhibition. Neuron 75, 490–502 (2012).

29. Sosnik, R., Haidarliu, S. & Ahissar, E. Temporal frequency of whisker movement.I. Representations in brain stem and thalamus. J. Neurophysiol. 86, 339–353(2001).

30. Rakic, P., Suner, I. & Williams, R. W. A novel cytoarchitectonic area inducedexperimentally within the primate visual cortex. Proc. Natl Acad. Sci. USA 88,2083–2087 (1991).

Supplementary Information is available in the online version of the paper.

Acknowledgements We thank S. Endo for the gift of the G-substrate antibody,N. Aagaard Jensen for the gift of the miR-Zbtb20 construct, F. Ango for the gift ofthe Sema3a-Tomato construct, and A. Goffinet for the gift of Dlx5/6::Cre/Celsr3flox

tissue. We are thankful to A. Benoit and F. Smets for technical assistance, to B. Goldingfor help with the in situ hybridizations, to O. Schaad, C. Barraclough and M. Docquier ofthe Genomics Platform of the University of Geneva for help with the microarrayexperiments. We thank E. Azim and F. Rijli for their comments on the manuscript. Workin the Jabaudon laboratory is supported by the Swiss National Science Foundation(SNF) (PP00P3_123447), theLeenaardsFoundation, theSynapsis Foundationand theNARSAD Foundation. C.B. was supported by an Ambizione grant from the SNF and bythe Gertrude von Meissner Foundation, C.L. by the SNF and the Simons FoundationAutism Research Initiative (SFARI), and F.C. and A.H. by the SNF (grant31003A-135631), EMBO, the International Foundation for Research inParaplegia, andthe Hans Wilsdorf Foundation.

Author Contributions D.J. and G.P. conceived the project, D.J. and C.B. designed theelectrophysiological experiments, and G.P., C.B., I.V., L.T., F.G. and D.J. performed theexperiments. D.J. and G.P. wrote the manuscript with help from C.L. and A.H.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of the paper. Correspondenceand requests for materials should be addressed to D.J. ([email protected]).

RESEARCH LETTER

4 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4



http://dx.doi.org/10.1111/ejn.12524


www.nature.com/reprints


mailto:[email protected]

METHODSMice. C57BL/6 and CD1 mice, Slc6a4Cre transgenic mice31, Ai14 transgenic reportermice (Jackson Laboratories, stock number 007914)32 and Rosa26::stopfloxDTA20 malesand females were used. Experiments were carried out in accordance with the Insti-tutional Animal Care and Use Committee of the University of Geneva and withpermission of the Geneva cantonal authorities.Histology. Postnatal mice were perfused with 4% paraformaldehyde (PFA) andbrains were fixed overnight in 4% PFA at 4 uC. 50-mm vibratome sections (LeicaVT1000S) were used for all histological experiments. S1L4 and S2L4 neurons werevisually identified on the basis of tangential location using topographical atlases(S1 versus S2) and bin location (L4 5 bins 3–4 out of 10, see Extended Data Fig. 2f, g).

In situ hybridization on slides was performed according to methods describedpreviously33. In brief, hybridization was carried out overnight at 60 uC with the dig-oxigenin (DIG)-labelled RNA probes. After hybridization, sections were washedand incubated with alkaline phosphatase-conjugated anti-DIG antibody (Roche1:2,000) overnight at 4 uC. After incubation, sections were washed and the colourreaction was carried out overnight at 4 uC in a solution containing NBT (nitro-bluetetrazolium chloride) and BCIP (5-bromo-4-chloro-39-indoly phosphate p-toluidinesalt) (Roche). After colour revelation, sections were washed, post-fixed for 30 minin 4% PFA and mounted with Fluromount (Sigma). For antisense probe synthesis,total complementary DNA was amplified by PCR with primers designed for speci-fic messenger RNA sequence of Pcdh20, NeuroD6 and Grm4. T7 or Sp6 promotersequence was added to the reverse primer sequence. DIG-labelled antisense RNAprobes were obtained after in vitro transcription of the resulted PCR product (Rochekit) or of the plasmid template for Rorb and Cdh8 (kind gift from M. Studer).

For immunohistochemistry, brain sections were incubated 1 h at room temper-ature in a blocking solution containing 3% BSA and 0.3% Triton X-100 in PBS andincubated overnight at 4 uC with primary antibodies: rabbit anti-5HTT (1/500, Cal-biochem, PC177L), rabbit anti-GSBS (1/2,000, gift from S. Endo)21, rabbit anti-CALB2(1/500, Swant, 7699/3H), guinea pig anti-VGLUT2 (1/2,000, Millipore, AB2251),mouse anti-GFAP (1/500, Sigma, G3893), rabbit anti-c-Fos (1/5,000, Santa Cruz,sc-52), rabbit anti-CUX1 (1:200, Santa Cruz, sc-13024). Sections were rinsed threetimes in PBS and incubated for 60–90 min at room temperature with the Alexa Fluor488- or 546-conjugated secondary antibodies (1/500, Invitrogen). For c-Fos immu-nostaining, biotinylated goat secondary anti-rabbit antibody (1/200, Invitrogen)were used, followed by an amplification step with Vectastain ABC kit (Vector Labo-ratories) and revealed in a 0.05% DAB, 0.02% NiSO4, 0.025% CoCl2 and 0.01% H2O2

solution. DAPI (0.2mg ml21) was used for fluorescent nuclear counterstaining.For Nissl staining, brain sections were mounted, stained with 0.5% cresyl violet

and dehydrated with graded alcohols. For cytochrome oxidase staining: free-floatingsections were placed in a solution of 0.5 mg ml21 DAB, 0.5 mg ml21 cytochrome C(Sigma), 40 mg ml21 sucrose, 0.1 mM Tris, pH 7.6 at 37 uC until staining appears.Fluoro-Jade B staining was performed as previously described34 with Fluoro-Jade Bkit (Millipore).Imaging and quantifications. All photomicrographs were taken with an Eclipse90i fluorescence microscope (Nikon, Japan) or with a LSM700 confocal microscope(Zeiss). Where cells were counted, this was performed on a minimum of three bio-logical replicates, within a 0.25 mm2 area encompassing the whole radial extent ofthe cortex, and divided into 5 or 10 bins. Photomicrographs are representative exam-ples taken from $2 replicates.Anterograde/retrograde labelling. Focal retrograde labelling from the cortex:anaesthetized P7 mice were placed on a stereotaxic apparatus thalamic neuronswere retrogradely labelled via 55-nl injections of FluoroGold (FG) 2% (Hydroxy-stilbamidine bis(methanesulphonate), Fluorochrome Inc.). Brains were collected2 days after the injection. Focal retrograde labelling from the Po (Extended DataFig. 9d): 92 nl of Alexa Fluor 555-conjugated cholera toxin subunit B (Invitrogen)was microinjected stereotactically at P10 and brains were collected at P23 andsectioned coronally. Only injections confined to the Po were analysed (n 5 4 Ctrland n 5 3 vb2 injections).

For anterograde labelling, the caudal part of fixed P7 brains was cut to reveal thecaudal thalamus. Trypan blue staining was applied to allow distinction between thal-amic nuclei and anterograde labelling was initiated by insertion of small NeuroVueDye-coated Filter Red (MTTI, FS-1002) (50 3 50mm2) in VB or Po. Brains wereincubated at 37 uC in 0.4% PFA for 8 weeks and cut into 100-mm sections on avibratome before immediate imaging. TC layer-specific arborization was quantifiedby measuring axonal length within each of the 5 bins using ImageJ software. Analysiswas performed on 3–5 100-mm-thick sections corresponding to bregma levels 20.82to 21.58 on the Paxinos adult brain atlas; S1 analyses were performed in the cortexat the same dorsolateral level as the hippocampus, whereas S2 was identified as theregion lateral to this, adjacent to the piriform cortex.Tissue microdissection and microarray. One mouse of control and vb2 litter-mates were used to collect each of three biological replicates of L2/3, L4, L5/6 sam-ples at P10 for cortical microarrays, or of Po and VB for thalamic microarrays.

Fresh coronal brain sections (140mm) were cut on a vibrating microtome andthalamic nuclei or cortical layers were visually identified and microdissected usinga Leica Stereomicroscope (Leica M165FC) in ice-cold oxygenated artificial cere-brospinal fluid under RNase-free conditions. Samples were stored in RNAlater at280 uC. RNA was extracted using an RNeasy kit (Qiagen) and one- or two-cycleamplification and labelling was performed (according to Affymetrix protocols) usingSuperscript cDNA synthesis kit (Invitrogen), MEGAscript T7 kit and MessageAmpIIaRNA amplification kit (Ambion). Labelled cRNA was fragmented and hybri-dized to Affymetrix Mouse Gene 1.0 ST arrays (for cortical samples) or mouse 4302.0 Genome arrays (for thalamic samples). GeneChips were incubated at 45 uCfor 16 h with biotin-labelled cRNA probes, and then washed and stained using astreptavidin-phycoerythrin conjugate with antibody amplification as described inAffymetrix protocol, using Affymetrix GeneChip Fluidics Station 450. GeneChipswere scanned on a GCS3000 scanner (Affymetrix).

Microarray CEL files were normalized using Robust Multichip Analysis andanalysed using Partek Genomics Suites software (http://www.partek.com). The accu-racy of the microdissection approach was validated by measuring concordance ofgene expression of the samples with genes confirmed to be strongly enriched in onecortical layer. Relative gene expression 5 (expression in the defined layer – meanexpression in all layers)/(mean expression in all layers)22. We defined the laminaridentity of L4, L2/3 and L5/6 neurons within S1 and S2 by using ANOVA (foldchange .2; P , 0.05, adjusted for multiple comparisons within Partek GenomicSuites) to identify their top 100 most specifically expressed genes in each of the sixcollected control samples. The relatedness between these six samples and L4vb2

was determined using unbiased cluster analysis of this reference set of genes (total600 transcripts, of which 453 are unique).In utero electroporation and tissue culture. E14.5 timed pregnant CD1 micewere anaesthetized with isoflurane/oxygen and a pCAG-GFP reporter plasmid orCMV-SEMA3A-Tomato35 (kind gift from F. Ango), pCAG-Npas4-GFP (ThermoFisher Scientific, MMM1013-202733015) or pCAG-miRZbtb20-GFP25 (kind giftfrom N. A. Jensen) construct was injected into the embryos’ lateral ventricle usinga 40-mm-tip glass micropipette mounted on a Nanoject II nanoinjector (DrummondScientific). E14.5 was chosen as a gestational age to specifically target L4 neuronprogenitors15. Voltage pulses (40 mV, 50 ms) were applied using external paddlesin order to target S1, as described previously15. Females were allowed to give birthand P7 pups were fixed with intracardiac perfusion of 4% PFA. Tangential sectionsof flattened electroporated cortex were cut at 50mm and immunostaining for VGLUT2was performed. Images were taken at LSM700 Zeiss confocal. The delineation betweenbarrel septae and hollows was performed on serially reconstructed image stacksusing the ImageJ software. The total number of primary dendrites and distributionof dendrites within hollows or septae were determined and statistically analysedusing an ANOVA test across conditions.

For in vitro culture, electroporated mouse embryos were dissected at E16.5 inice-cold HBSS. The S1-positive electroporated site was microdissected under fluo-rescent microscopy and dissociated cells cultured during 3 days on 12-mm, coatedcoverslips (poly-L-lysine, 0.1 mg ml21) in 500ml Neurobasal supplemented withGlutaMAX, B27, sodium pyruvate and a mixture of antibiotic penicillin-streptomycinall from Gibco. The cultures were then fixed with 4% PFA, mounted with DAPIand analysed on a LSM 700 Zeiss confocal and ImageJ software.In vitro electrophysiology. AAV-mediated expression of ChR2 (AAV5-hsyn-hChR2 H134R) was injected at P0 in the Po, under ultrasound guidance (Vevo 660,VisualSonics). Injected mice were collected 3 weeks later and processed for elec-trophysiology. ,P23 mice were deeply anaesthetized with isoflurane and were thendecapitated. Brains were removed and placed in cold (0–4 uC), oxygenated (95%O2–5% CO2) slicing solution containing: 119 mM NaCl, 2.5 mM KCl, 1.3 mM MgCl,2.5 mM CaCl2, 1.0 mM Na2HPO4, 26.2 mM NaHCO3 and 11 mM glucose. Coronalslices (300 mM) were kept at room temperature and were allowed to recover forat least 1 h before recording. Under low magnification, the barrels in L4 could bereadily identified, and high-power magnification was used to guide the recordingelectrode onto visually identified neurons. The radial extent of the cortex was vir-tually divided into 5 bins (1 5 most pial) and patched neurons were always locatedin bin 2, which corresponds to L4, as validated in pilot experiments in which neu-rons were filled with biocytin (3 mg ml21) for 15 min and location was assessedusing immunostaining of CUX1 after fixing fresh sections (n 5 7 Ctrl and n 5 6vb2) in both S1 and S2. The internal solution contained 140 mM potassium glu-conate, 5 mM KCl, 10 mM HEPES, 0.2 mM EGTA, 2 mM MgCl2, 4 mM Na2ATP,0.3 mM Na3GTP and 10 mM sodium creatine-phosphate. Currents were ampli-fied (Multiclamp 700B, Axon Instruments), filtered at 5 kHz and digitized at 20 kHz(National Instruments Board PCI-MIO-16E4, Igor, WaveMetrics). The liquid junc-tion potential was 112 mV. Experiments were discarded if the access resistancevaried by more than 20%. Synaptic currents were evoked by light flashes deliveredby a fibre optic cable (Thorlabs) attached to a 473-mm solid-state laser (CrystaLaser).Cells were held at 260 mV and GABA currents were blocked by wash in picrotoxin

LETTER RESEARCH


http://www.partek.com

(Tocris, 100mM). Light-evoked EPSCs were recorded at 246 mV and light-evokedIPSCs were recorded at 0 mV (ref. 36) (Extended Data Fig. 8).In vivo electrophysiologyIntrinsic optical imaging. ,P23 mice were first anaesthetized with isoflurane(4% for induction with ,0.5 l min21 O2), and then with urethane (1.5 g kg21, i.p.,prepared in lactated ringer solution containing in mM: 102 NaCl, 28 Na L-Lactate,4 KCl, 1.5 CaCl2). Eye ointment was applied to prevent dehydration. The scalp waslocally anaesthetized with lidocaine (1%), the periosteum gently removed, and acustom-made plastic chamber was attached to the skull above barrel cortex (centred1.5–2 mm posterior from bregma, 2–2.5 mm lateral) with dental acrylic and dentalcement. The chamber was filled with sterile cortex buffer (containing in mM: 125NaCl, 5 KCl, 10 glucose, 10 HEPES, 2 CaCl2, and 2 MgSO4, pH 7.4) and sealed witha glass coverslip. Intrinsic optical signals were imaged through the intact skull usingan Imager 3001F28.In vivo whole-cell patch clamp. After imaging, a small, ,1 3 1 mm piece of bonewas removed using a dental drill (centred above the C2 whisker maximum intrinsicoptical signal response). Whole-cell ‘blind’ patch-clamp recordings were obtainedas previously described28. High positive pressure (200–300 mbar) was applied tothe pipette (7–9 MV) to prevent tip occlusion while penetrating the dura. Afterpassing the dura the positive pressure was immediately reduced to prevent corticaldamage. The pipette was then advanced in 2-mm steps, and pipette resistance wasmonitored in the conventional voltage-clamp configuration. When the pipette resis-tance suddenly increased, positive pressure was relieved to obtain a 3–5-GV seal.After break-in, the membrane potential (Vm) was measured, and dialysis was allowedto occur for at least 5 min before deflecting the whisker. Data were acquired using aMulticlamp700B amplifier (Molecular Devices), and digitized at 10 kHz (NationalInstruments), using MATLAB (Mathworks)-based Ephus software (http://research.janelia.org/labs/display/ephus; The Janelia Farm Research Center). Off-line analysiswas performed using custom routines written in IgorPro (Wavemetrics). All neu-rons were located at ,200mm below the pia.

Current-clamp recordings were made using a potassium-based internal solution(in mM: 135 potassium gluconate, 4 KCl, 10 HEPES, 10 Na2-phosphocreatine,4 Mg-ATP, 0.3 Na-GTP, pH adjusted to 7.25 with KOH, 285 mOsm). Series resis-tance (Rs) and input resistance (Rin, not including Rs) were monitored with a 100-mslong-lasting hyperpolarizing square pulse 400 ms before each whisker deflection,and extracted off-line by using a double exponential fit. Recordings were discardedif one of the following conditions occurred: (1) Vm and Rs exceeded 250 mV and50 MV, respectively; (2) spontaneously occurring spikes were not overshooting;(3) Rs or Rin changed more than 30% over the duration of the experiment. Thebridge was usually not balanced and liquid junction potential was not corrected.Whisker-evoked post-synaptic potential (PSP) analysis. Whisker-evoked PSPswere evoked by forth and back deflection of the whisker (100 ms, 0.1 Hz) usingpiezoelectric ceramic elements attached to a glass pipette ,4 mm away from theskin. The voltage applied to the ceramic was set to evoke a whisker displacement of,0.6 mm with a ramp of 7–8 ms. The C1 and C2 whiskers were independentlydeflected by different piezoelectric elements. PSP analyses were confined to downstates. Peak amplitude and integral analysis was performed on each trace, and thenpresented as a mean of at least 30 whisker-evoked responses. The PSP onset latencywas defined as the time point at which the amplitude exceeded 3 3 s.d. of the base-line noise over 5 ms before stimulation. The jitter was defined as the standard devi-ation of at least 30 whisker-evoked responses (control mice, n 5 7 cells per 3 mice;vb2 mice, n 5 6 cells per 3 mice).Environmental enrichment. All but one whisker (C1) on the right side of thesnout were clipped in P23 vb2 or control mice (n 5 3), after which the animalswere placed in an large playground box containing plastic balls, maze-like pieces ofthread, and various small objects, as previously described. Mice were kept for 1 h inenriched environment, perfused and brains stained for c-Fos expression14,27,37.Noxious stimulation. P23 vb2 or control mice were briefly anaesthetized withisoflurane, and 50ml of a capsaicin solution (10 mM (Sigma), 100% ethanol and 7%Tween-80 in saline) was subcutaneously injected in the whisker pad as previouslydescribed14,38. Mice were allowed to wake up and were euthanized after 1 h and theirbrain stained for c-Fos expression. The number of c-Fos1 neurons was quantifiedon five sections per animal (n 5 3) in a ,600-mm width of S1 (,3 barrels). Statis-tical comparisons of layer-specific c-Fos activation between control and vb2 micewere done using Student’s t-test.Behavioural analysis. ,P30 vb2 mice and littermates were used for all behavi-oural experiments. Both males and females were used. To habituate animals to thetesting environment, mice were transported within their home cage to the testingroom for three consecutive days before testing. All behavioural tests took place duringthe light phase of the light–dark cycle, with the observer blinded to the mouse genotype.

For the open field test, spontaneous locomotor activity was examined in n 5 6vb2 mice and n 5 17 control littermates. Mice were placed in the centre of a 453 45-cm

arena and locomotion in the dark (distance travelled, percentage time spent incentral, intermediate or lateral sectors) was recorded during 15 min using an infraredcamera coupled to a tracking software (ANYmaze, Stoelting Co). For the grid-walktest, sensorimotor abilities (limb placement accuracy, coordination) of hindlimbswere examined in n 5 4 vb2 mice and n 5 13 control littermates by assessing theaptitude of the mice to navigate in dim lighting over a wire mesh grid (2.4 3 2.4-cmgrid spaces, 45 3 45-cm total area) during 6 min39,40. Foot-faults were counted whena hindlimb paw protruded entirely through the grid. For the gap-crossing test,whisker-sensing abilities and sensorimotor coordination were examined in n 5 7vb2 mice and n 5 6 control littermates by placing mice on a 12 3 12 cm elevatedplatform and measuring the number of crosses performed to a neighbouring plat-form across a 4-cm gap in obscurity. The gap was such that the mouse was required toextend its head and detect the opposite platform with its whiskers before crossing41,42.The number of crosses and failed attempts (that is, falls) were recorded by an observervia an infrared camera. A maximum of two falls per mouse was admitted afterwhich the test was interrupted. Adhesive patch removal task: this test was originallydeveloped to assess somatosensory asymmetry and sensory function after sensor-imotor cortex lesions43–45. n 5 5 vb2 mice and n 5 12 control littermates were used.A 6-mm diameter circular adhesive patch was placed on the plantar surface of eachhindpaw, after which mice were released in the testing arena and observed for 240 s.Latency to detect the first patch (snout contact with the patch) as well as the timetaken to remove both patches was measured. Mice underwent three consecutivetrials, with an inter-trial interval of 60 min, and values were averaged for each mouse.For the tail-flick test, sensitivity to noxious stimuli was determined using the tail-flick test46 in n 5 5 vb2 mice and n 5 4 control littermates. In the apparatus (IITCLife Science), a pre-focused light beam supplied an area of 4 3 6-mm heat stimu-lation to the tail. The time taken for the mouse to flick its tail away from the stim-ulation area was recorded, providing a measure of pain sensitivity. Values wereobtained from three sessions of ten trials (inter-session interval: 24 h).Statistics. No statistics were used to determine group sample size; however, sam-ple sizes were similar to those used in previous publications from our group andothers. The person performing the test was blinded to the animal’s genotype. Allmice were used in the study; the tail-flick test was performed only if the animalcollaborated to rest with its tail in the flick detection groove. If animals undertookmore than one task, the order of the task was randomized. Two-tailed t-tests wereused for all statistical analyses except for the tail-flick test, for which values had anon-Gaussian distribution, which was analysed using a Kolmogorov–Smirnov test.Values are shown as mean 6 s.e.m. throughout the manuscript. n values refer tobiological replicates throughout the manuscript.

31. Zhuang, X., Masson, J., Gingrich, J. A., Rayport, S. & Hen, R. Targeted geneexpression in dopamine and serotonin neurons of the mouse brain. J. Neurosci.Methods 143, 27–32 (2005).

32. Madisen, L. et al. A robust and high-throughput Cre reporting and characterizationsystem for the whole mouse brain. Nature Neurosci. 13, 133–140 (2010).

33. Lai, T. et al.SOX5 controls the sequential generation of distinct corticofugal neuronsubtypes. Neuron 57, 232–247 (2008).

34. Schmued, L.C.&Hopkins,K. J. Fluoro-JadeB:a highaffinity fluorescent marker forthe localization of neuronal degeneration. Brain Res. 874, 123–130 (2000).

35. Cioni, J.-M. et al. SEMA3A signaling controls layer-specific interneuron branchingin the cerebellum. Curr. Biol. 23, 850–861 (2013).

36. Gabernet, L., Jadhav, S. P., Feldman, D. E., Carandini, M. & Scanziani, M.Somatosensory integration controlled by dynamic thalamocortical feed-forwardinhibition. Neuron 48, 315–327 (2005).

37. Bisler, S. et al. Expression of c-Fos, ICER, Krox-24 and JunB in the whisker-to-barrelpathway of rats: time course of induction upon whisker stimulation by tactileexploration of an enriched environment. J. Chem. Neuroanat. 23, 187–198 (2002).

38. Noma, N. et al. Organization of pERK-immunoreactive cells in trigeminal spinalnucleus caudalis and upper cervical cord following capsaicin injection into oraland craniofacial regions in rats. J. Comp. Neurol. 507, 1428–1440 (2008).

39. Z’Graggen, W. J., Metz, G. A., Kartje, G. L., Thallmair, M. & Schwab, M. E. Functionalrecovery and enhanced corticofugal plasticity after unilateral pyramidal tractlesion and blockade of myelin-associated neurite growth inhibitors in adult rats.J. Neurosci. 18, 4744–4757 (1998).

40. Leingartner, A. et al. Cortical area size dictates performance at modality-specificbehaviors. Proc. Natl Acad. Sci. USA 104, 4153–4158 (2007).

41. Hutson, K. A. & Masterton, R. B. The sensory contribution of a single vibrissa’scortical barrel. J. Neurophysiol. 56, 1196–1223 (1986).

42. Jenkinson, E. W. & Glickstein, M. Whiskers, barrels, and cortical efferent pathwaysin gap crossing by rats. J. Neurophysiol. 84, 1781–1789 (2000).

43. Hernandez, T. D. & Schallert, T. Seizures and recovery from experimental braindamage. Exp. Neurol. 102, 318–324 (1988).

44. Thallmair, M. et al. Neurite growth inhibitors restrict plasticity and functionalrecovery following corticospinal tract lesions. Nature Neurosci. 1, 124–131 (1998).

45. Tomassy, G. S. et al. Area-specific temporal control of corticospinal motor neurondifferentiation by COUP-TFI. Proc. Natl Acad. Sci. USA 107, 3576–3581 (2010).

46. Le Bars, D., Gozariu, M. & Cadden, S. W. Animal models of nociception. Pharmacol.Rev. 53, 597–652 (2001).

RESEARCH LETTER


http://research.janelia.org/labs/display/ephus

http://research.janelia.org/labs/display/ephus

Extended Data Figure 1 | Early postnatal VB ablation in vb2 mice.a–f, Coronal sections of the thalamus and S1 cortex at P0 (a, b), P4 (c, d) and P7(e, f) in control (a, c, e) and vb2 (b, d, f) mice. Staining for Fluoro-Jade B (FJ)reveals a peak of VB-specific neuronal degeneration at P4 in vb2 mice, atime at which the VB-specific marker GSBS21 is already lacking. VGLUT2immunostaining indicates that TC axons are present in deep cortical layers ofvb2 mice at P0 (arrowhead) and reach L4 at P4, a time at which VBdegeneration is at its peak. These axons are not clustered into whisker-relatedbarrels and do not express the VB-specific marker GSBS, which is normallyexpressed at P4. DAPI (49,6-diamidino-2-phenylindole) staining indicates thatpostsynaptic L4 neurons fail to assemble into barrels in vb2 cortex. At P7,

immunostaining for the glial marker GFAP shows a secondary glial ‘scar’ (redarrowheads) at the ventral thalamic border in vb2 mice (e, f). g, Quantificationof FJ1 neurons. h, i, Schematic representation of VB (blue) and Po (red)axonal development in control (h) and vb2 (i) S1 cortex. j, k, Nissl (j) andcytochrome oxidase (k) stainings reveal specific ablation of the VB in vb2 mice.l, In situ hybridization for the interneuron marker GAD67 shows preserveddelineation of the dorsal and the ventral thalamus (red arrowheads).m, Schematic representation of vb2 thalamus. Scale bars, 200mm (insets,50mm). CP, cerebral peduncle; CP, cortical plate; dLG, dorsolateral geniculatenucleus; LD, laterodorsal nucleus; LP, lateroposterior nucleus; PF,parafascicular nucleus; VM, ventromedial nucleus.

LETTER RESEARCH


Extended Data Figure 2 | Rewiring of input to S1L4 occurs in the absenceof cortical neuron death. a, Nissl-stained flattened preparation of thesomatosensory cortex showing lack of barrel-like clusters in postsynaptic L4cortical neurons in vb2 mice (empty red arrowhead). b, c, Presynaptic TCterminals are still present in L4 of vb2 cortex at P23 using the pan-TCpresynaptic marker VGLUT2 (white arrowheads) (b) whereas cortical barrelsare lacking (c). d, e, Fluoro-Jade B, cleaved caspase 3 (CASP3) (d) and GFAP

(e) expression do not show evidence of cortical neuron degeneration in vb2

mice (inset shows non-cortical CASP31 neurons in a non-cortical region ofthe same section). f, g, Staining and quantification using DAPI (f) and theL2/3-L4 marker CUX1 (g) show lack of barrels but preservation of S1L4 cellnumbers in vb2 mice. Total quantification surface: 0.25 mm2. Scale bars,400mm (a), 200 mm (b–d, f, g), inset 10mm (d).

RESEARCH LETTER


Extended Data Figure 3 | Ectopic TC projections to S1 in vb2 mice originatefrom the Po. a–c, In control mice, retrograde labelling from S1 usingFluoroGold (FG) labels both GSBS1 VB neurons and CALB21 Po neurons(b, numbered blue arrowheads), whereas other thalamic nuclei are notdetectably labelled (b, c). d–f’, In vb2 mice, retrograde labelling from S1exclusively labels GSBS2 Po neurons (d, blue arrowhead). No additionallabelling is found in other thalamic nuclei (d, e). FG1 Po neurons are CALB21

and located outside of the glial scar (f, f’: high magnification from inset in f).g, Povb2 neurons were undistinguishable from control Po neurons bymolecular and microarray comparative gene expression analysis between

Povb2, Po and VB neurons demonstrating that they are bona fide Po neurons.Heatmap representation of the expression intensity of the 100 most VB-specificgenes in VB, Po and Povb2 neurons. None of these genes are statisticallysignificantly upregulated in Povb2 neurons compared to Po. h, i, Po-S2L4connectivity is normal in vb2 as assessed by anterograde labelling of Poprojections in S2L4vb2 cortex (h) and S2L4 neuron responses to optogeneticstimulation of Po axons (i). Scale bars, 1 mm (low-magnification images)(a–f) and 100mm elsewhere. AM, anteromedial; AV, anteroventral; Cx, cortex;dLG, dorsolateral geniculate; Hip, hippocampus; LD, laterodorsal; LP,lateroposterior; PF, parafascicular; VL, ventrolateral; VM, ventromedial.

LETTER RESEARCH


Extended Data Figure 4 | VBRPo switch in input leads to downregulationof S1L4 transcripts and upregulation of S2L4 transcripts in S1L4vb2

neurons. a, In situ hybridizations (expression density map) from the AllenBrain Atlas (ABA) database showing expression of three sample genesspecifically expressed in L2/3, L4 and L5 (Mdga1, Rorb, Fezf2, respectively).The heatmap on the right represents relative expression intensity in themicroarray samples, which is concordant with the ABA data. Relativeexpression 5 (expression in the defined region – mean expression in all layers)/

(mean expression in all layers) (see ref. 22). b, Heatmap representingsample-specific gene expression for the union of the top 100 most specificallyexpressed genes of L4vb2, L2/3, L4 and L5/6 samples in S1 and S2. Note thatS1L4vb2 gene expression is intermediate between S1L4 and S2L4 neurons.c–e, In situ hybridization showing downregulation of the S1L4-enrichedtranscripts Pcdh20 (c) and Grm4 (d), and upregulation of S2L4-enrichedtranscripts NeuroD6 (e) in S1L4vb2 neurons. Scale bars, 1 mm (lowmagnifications), 100mm (high magnifications), 30mm (inset).

RESEARCH LETTER


Extended Data Figure 5 | S1L4vb2 developmental gene expression isS2L4-like. a, Schematic representation of VB (blue) and Po (red) axonaldevelopment in control S1 (left), control S2 (middle) and S1vb2 (right) cortex.Boxed area indicates region shown in c. b, Summary of the findings: the timecourse of expression of L4 gene expression in S1L4vb2 neurons is similar to

that of S2L4 control cortex. Values are colour-coded using S1L4 P0 controlvalues as baseline. c, In situ hybridizations for Rorb and Grm4 (S1L4-type) andCdh8 and NeuroD6 (S2-L4-type) transcripts at P0, P4 and P10 indicate thatS1L4vb2 developmental gene expression is S2L4-like. Scale bars, 100mm.CP, cortical plate. See also Supplementary Note 2.

LETTER RESEARCH


Extended Data Figure 6 | Loss of VB input and acquisition of Po input eachdefine genetic changes in S1L4vb2 neurons. a, Schematic representation ofthe phenotypes examined. Celsr3 conditional knockouts (cKO) S1L4 neuronslack both VB and Po inputs (see Supplementary Note 3) whereas S1L4vb2

receive Po but not VB input. b, Summary of the findings: expression of Rorband Pcdh20 expression is decreased both in Celsr3 cKO and vb2 L4 neurons,but this decrease is mitigated by Po input in vb2 cortex. By contrast, Grm4

expression is not rescued by Po input and is thus VB-dependent. Cdh8upregulation in vb2 L4 neurons depends on Po input as it does not occur inthe absence of TC input (Celsr3 cKO). c, In situ hybridizations showingexpression of the S1L4-type genes Rorb, Pcdh20 and Grm4, and the S2L4-typegene Cdh8. The two photomicrographs with an asterisk are also presented inExtended Data Fig. 4. Normalized intensity values were obtained by radialscanning of intensity using the gel tool of ImageJ software. Scale bars, 200mm.

RESEARCH LETTER


Extended Data Figure 7 | VB input regulates expression of genescontrolling neurite differentiation and polarity in S1L4 neurons. a, Npas4and Sema3a expression is increased in vb2 mice, whereas Zbtb20 is decreased.Open circles indicate values for individual replicates, value within bars indicateP value obtained with the microarray analysis. b, Overexpression of Npas4and Sema3a or downregulation of Zbtb20 using a miR construct in S1L4neurons collected 2 days after in utero electroporation at E14.5 led to changes incell polarity. Dendrites are preferentially oriented away from the axon (yellowarrowhead, 0u in bulls eye plot), whereas they are evenly distributed when a

controlGFP plasmid is used. n and P values are indicated in the figure, ANOVA.c, In vivo overexpression of Npas4, Sema3a or miR-Zbtb20 by in uteroelectroporation at E14.5, the time of birth of L4 neurons, impairs dendriticorientation towards VB axons located in barrel hollows at P7 (VGLUT21

region within barrels, pink area in single-cell displays) and increases in thenumber of primary dendrites for Npas4 and miR-Zbtb20 (*P , 0.05, ANOVA,n values indicated within bars). Septae correspond to spaces between barrels(grey area in single cell displays). Scale bars, 10mm (b), 20mm (c). See alsoSupplementary Note 4. Values: mean 6 s.e.m.

LETTER RESEARCH


Extended Data Figure 8 | S1L4vb2 E/I input balance is S2L4-like.a, Schematic summary of the experiment: S1L4, S2L4 and S1L4vb2 neuronswere recorded while optogenetically stimulating the VB (for S1L4 neurons) orthe Po (for S2L4 and S1L4vb2 neurons). Feed-forward inhibitory input onto L4neurons was determined by changing the holding potential as detailed in

Methods. b–d, Sample traces (b) normalized to the excitatory input amplitude(c) and average values (d) showing that E/I ratios are increased to S2L4 levels inS1L4vb2 neurons. Values: mean 6 s.e.m. *P , 0.05, ANOVA; ns, notsignificant (P 5 0.4).

RESEARCH LETTER


Extended Data Figure 9 | Pre-thalamic trigeminal input pathways are notrewired in vb2 mice. a, Schematic representation of trigeminothalamicpathways and summary of the findings. Input to the VB, which conveysinformation on whisker contacts, originates from the PrV nucleus of thetrigeminal complex, forming the lemniscal pathway, with a small contingent offibres reaching the VB–Po border, see Supplementary Note 6 for details). Inputto the Po, which forms the paralemniscal pathway, originates from the SpVinucleus (interpolaris part of the spinal nucleus) of the trigeminal complex.In vb2 mice, the PrV nucleus is markedly atrophied, presumably owing to loss

of VB targets, and only a few cells subside. The paralemniscal pathway isunaffected. Dashed lines indicate location of the sections shown in b–d.b, Cytochrome oxidase staining showing markedly atrophied PrV in vb2

mice. c, The PrV is not detectably activated by whisker contacts duringenvironmental exploration in vb2 mice, whereas the SpVi is unaffected.d, Retrograde labelling from the Po shows numerous labelled neurons in theSpVi and sparse labelled neurons in both control and vb2 mice (n 5 5 Ctrl andn 5 3 vb2 injections). Scale bars, 100mm.

LETTER RESEARCH


Extended Data Table 1 | VB input regulates a core set of genes involved in neurite differentiation and polarization

RESEARCH LETTER


W W W. N A T U R E . C O M / N A T U R E | 1

SUPPLEMENTARY INFORMATIONdoi:10.1038/nature13390

Pouchelon et al., Supplementary Notes, p.1

Pouchelon et al. SUPPLEMENTARY NOTES

Supplementary Note 1. Rewiring of Po input onto S1L4 neurons is limited to a critical period of development and normal Po projections to S2L4 neurons are unaffected in vb- mice. Using quinolinic acid to perform focal ablation of VB neurons in adult mice, Croquelois and collaborators have reported that deafferented S1 cortex does not receive novel TC input47. Confirming these findings, when we focally ablated VB neurons at P10 using chemical or genetic (conditional ablation using diphtheria toxin receptor (DTR)-expressing lentivirus48) approaches, we did not observe rewiring of Po axons into L4 (data not shown). Therefore, extensive Po rewiring and changes in S1L4 neuron identity do not occur in more mature circuits. There is therefore a “critical period” during which rewiring and identity changes can occur, which is consistent with previous findings from our laboratory in which direct reprogramming of L4 neurons with targeted expression of Fezf2 was only possible during the first postnatal week15.

In vb− mice, Po-evoked synaptic responses of S2L4 neurons by optogenetic stimulation of Po axons, as well as the anatomical projections of Po neurons to these neurons were normal indicating that Po axon rewiring does not occur at the expense of normal Po targeting (Extended Data Fig. 3h,i).

Supplementary Note 2. Developmental time course of genetic changes in S1L4vb-. The developmental time course of genetic changes in S1L4vb- neurons was examined by comparing expression of select VB input-dependent genes at P0, P4, and P10 to that of S1L4 and S2L4 neurons of control and vb- mice. (Extended Data Fig. 5). The expression of S1L4-type genes Rorb and Grm4 was only weakly increased in S1L4vb- neurons, while the expression of S2L4-type genes Cdh8 and NeuroD6 was upregulated compared to S1L4, as was the case during normal development of S2L4 neurons. Therefore, expression of S1L4 VB input-dependent genes identified here is developmentally regulated, and their time course of expression coincides with the maturation of VB-L4 synapses and barrel formation (Extended Data Fig. 1a-i). In S1L4vb- neurons, this VB input-dependent gene expression maturation does not occur, resulting in an S2L4-type developmental gene expression program.

Supplementary Note 3. Loss of VB input and acquisition of Po input both instruct the differentiation programs of S1L4vb- neurons. In order to identify the specific contribution of loss of VB input and acquisition of Po input to the gene expression changes in S1L4vb- neurons, we examined gene expression in the absence of all TC input, using Dlx5/6::Cre/Celsr3flox mice (Celsr3 cKO), in which TC axons get misrouted and never reach the cortex due to conditional ablation of Celsr3 in the subpallium23. By comparing S1L4 gene expression in vb- mutants (which receive only Po input) and in Celsr3 cKO mutants (which receive no TC input), we identified the contribution of Po input to S1L4vb- genetic changes.

Our results indicate that Po rewiring directly contributes to S1L4 transcriptional programs (Extended Data Fig. 6), acting on some genes to mitigate the loss of VB input (Rorb, Pcdh20, where Ctrl expression > vb− > Celr3 cKO), on others to instruct expression of S2L4-type genes (Cdh8, where vb− > Celsr3 = Ctrl), while having no effect on the expression of some strictly VB input-dependent genes (Grm4, where Ctrl > Celsr3 cKO = vb−). The overall genetic identity of S1L4vb- neurons thus reflects the combined effects of loss of VB input-dependent genes, partial rescue of this loss by Po afferentation and acquisition of Po-dependent genes, whose net effect is the acquisition of an S2L4-like genetic identity. Together, these results indicate that VB and Po TC axons each instruct the expression of distinct, partially overlapping sets of genes in S1L4vb- neurons.

Supplementary Note 4. Functional relevance of VB input-dependent genes. Many of the genes presented in Extended Data Table 1 have been shown to be critical for VB circuit assembly and cortical barrel formation. These include several transcription factors such as Btbd3 (↓1.5-fold in vb−), which controls L4 neuron dendritic orientation towards TC axons49, Bhlhb5 (↓1.5-fold) and Lmo4 (↑1.9-fold), which each are required for the clustering of L4 neurons into barrels50,51 and Rorb (↓2.1-fold), which controls the patterned clustering of cortical neurons18, as well as genes coding for extracellular proteins such as Vgf (↑2.4-fold) and Nrn1 (↑2.9-fold), which promote neurite outgrowth in L4 cortical neurons and is also released by TC axons52.

The three novel VB input-dependent transcripts characterized here (Fig. 3j and Extended Data Fig. 7), include two transcription factors, Zbtb20 (↓1.5-fold), which is a transcriptional repressor (including for targets such as Rorb, Cux2, and Fezf2)53 and has been shown to control dendritic complexity in hippocampal neurons25, and Npas4 (↑2.2-fold), which regulates neurite outgrowth24 and the number and distribution of inhibitory synapses made onto excitatory hippocampal neurons in an activity-dependent manner54,55, and an extracellular protein, Sema3a (↑1.6-fold), which regulates neuronal polarization in hippocampal and cortical neurons26.

Using in vitro cultures of S1L4 neurons overexpressing Npas4 or Sema3a or in which Zbtb20 expression was downregulated using a miR construct25, we found that these three transcripts cell-autonomously regulate the polarity of S1L4 neurons, as indicated by a greater proportion of neurites extending 180o away from the axon (Extended Data Fig. 7b). We further extended these results in vivo using in utero electroporation at E14.5, the time of birth of L4 neurons, to show that each of these genes impairs the ability of S1L4 neurons to polarize their dendrites towards the center of barrels (hollows), where VB axons are located, thereby compromising the postnatal columnar assembly of cortical circuit and associated topographic representation of vibrissal input (Extended Data Fig. 7c).

Together, these data indicate that VB input regulates a core set of genes critical for S1L4 neuron dendritic polarization and associated columnar assembly of cortical neurons into circuits. This transcriptional regulation involves activation as well as repression of functional target genes, indicating that both processes are at play during input-dependent differentiation.

Supplementary Note 5. Trigeminal input to the thalamus is not rewired in vb- mice. Vibrissal input to the VB is normally provided by the principal nucleus of the trigeminal complex of the brainstem (PrV) via the lemniscal pathway56 (Extended Data Fig. 9a, in blue). In addition, a minor contingent of PrV neurons have dual projections to both the VB and the VB-Po border, forming the “dorsal pathway”57. Input to the Po on the other hand mostly originates from neurons in the rostral interpolaris part of spinal nucleus of the trigeminal complex (SpVi), whose axons form the paralemniscal pathway (Extended Data Fig. 9a, in red).

We examined the anatomy, function, and connectivity of PrV and SpVi in vb− mice, and found that PrV output is reduced to a minimal number of residual neurons, while SpV output (i.e. paralemniscal) is morphologically and functionally unchanged, providing “normal” input to a cortically-rewired Po nucleus. Therefore the locus of the plasticity reported in this study is largely thalamocortical, i.e. confined to the rewiring of Po input onto S1L4 neurons.

a- Anatomical assessment Anatomical examination of the PrV using cytochrome oxidase staining revealed a striking atrophy of this nucleus with loss of barrelette patterning, while the SpVi nucleus was unchanged

SUPPLEMENTARY INFORMATION

2 | W W W. N A T U R E . C O M / N A T U R E

RESEARCH


compared to control mice (Extended Data Fig. 9b). This suggests that most PrV neurons degenerate following loss of their target (the VB) in vb− mice. b- Functional assessment We examined activation of the distinct trigeminal nuclei in control and vb− mice during environmental exploration. This paradigm normally strongly and specifically activates neurons of the PrV and lemniscal pathway using c-Fos as a reporter for activity, as we have recently shown14. As expected from the atrophic PrV, only sparse activated neurons were found in this nucleus in vb− mice, while activation of the SpVi was similar to control mice (Extended Data Fig.9c). This indicates that PrV atrophy and associated loss of lemniscal pathway input does not lead to a detectable compensatory increase in paralemniscal pathway activity during environmental exploration. c- Assessment of connectivity We investigated whether VB ablation led to a rewiring of PrV neurons onto Po neurons by retrogradely labeling trigeminal Po afferents using stereotaxic microinjections of labeled Cholera Toxin B subunit (CTB) into the Po at P10. Consistent with the anatomical and functional data above, when examined at P23, retrogradely-labeled neurons were mostly located in the SpVi, while only sparse labeled neurons were present in the PrV in both control and vb- mice (Extended Data Fig. 9d). Therefore, early postnatal VB ablation does not lead to a significant rewiring of lemniscal PrV neurons into the Po.

Supplementary Note 6: Expanded topographical activation of S1L4vb- neurons. VB neurons normally receive input from single whiskers and have topographically constrained projections to L4 neurons in their target whisker-related barrel. In contrast, Po neurons normally receive input from multiple whiskers and have more diffuse projections, which target multiple adjacent barrel columns14,56. Previous studies have reported that functional barrel columns (i.e. the topographical distribution of cortical neurons activated by single whisker) are disproportionally enlarged and diffuse in mutants where the convergence of VB projections are impaired (e.g. Ebf1 cKO58) or when the distribution of postsynaptic cortical neurons is disrupted (e.g. Reeler mouse27).

Several factors may account for the lateral diffusion in the columnar cortical activation in vb- mice in response to whisker contacts. First, the actual distribution of disrupted TC projections may be more widespread than appears using anatomical tracing, since thin long-distance processes may not always be identified. Second, disrupted TC input may induce secondary changes in intracortical connectivity, leading to de-repression of latent intracortical pathways, as suggested by “S2-like” increases in excitation/inhibition ratios in response to Po stimulation (Extended Data Fig. 8).

The fact that S1L4vb- excitation/inhibition ratios become “S2-like”, together with overlap between the laminar distribution of activated neurons and the anatomical distribution of Po axons suggests that de-repression of these pathways, if occurring, is a specific process related to acquisition of Po input. This increased activation is not present in baseline conditions, but instead is evoked by whisker pad stimulation (see Fig. 4b, activated vs. control side in vb− cortex), and therefore reflects a genuine input-dependent change in cortical activity.

Therefore, type-dependent controls over the area-specific assembly of intracortical excitatory and inhibitory circuits could contribute to the “S2-like” lateral diffusion of cortical activation upon VB→Po rewiring.

Supplementary Note 7: vb- mice have decreased haptic skills and increased sensitivity to noxious stimuli. As a first step to investigate sensorimotor function in vb− mice, we performed an open-field exploration test, in which mice are allowed to freely explore their environment (Supplementary Video 1). The

behavior of vb− mice was indistinguishable from that of control mice in this test, indicating that loss of VB (which relays somatosensory input from the whole body) does not lead to detectable deficits in locomotor and exploratory activities.

However, in a more complex sensorimotor task, the grid walk test39,40, in which mice have to walk over a 2.4 cm-spaced wire mesh grid, vb− mice made significantly more foot faults than littermate controls (Fig. 4e, P <0.01, n = 13 Ctrl, n = 4 vb− mice, Student’s t-test). Similarly vb- mice were more likely to fall when crossing from one platform to another across a 4 cm gap in the dark, a test that has been used to assess function of the whisker-to-barrel pathway (Fig. 4e, P < 0.05, n = 7 Ctrl, n = 6 vb− mice, Student’s t-test)41,42. Therefore, disrupted activation of S1L4 neurons to tactile stimuli is associated with impaired sensorimotor function in vb− mice.

In order to determine the contribution of somatosensory impairment to these deficits, we performed an adhesive patch removal test, in which the mouse has to remove an adhesive patch placed on each hindpaw. This test was originally developed to assess sensory function after sensorimotor cortex lesions and has been shown to be particularly sensitive to somatosensory function43-45. Consistent with a pronounced impairment in somatosensation, vb− mice were much less efficient at detecting the patch (first contact latency) and removing it (Fig. 4e, P = 0.008 for detection, P < 0.0005 for removal, n = 12 Ctrl, n = 5 vb− mice Student’s t-test), suggesting that impaired coordination in vb- mice has a significant somatosensory component.

We next examined the behavioral responses of vb− mice to noxious stimulation. Compared to control mice, vb− mice displayed increased and prolonged (up to 20 minutes) behavioral signs of pain following administration of capsaicin in the whisker pad, including immobility, panting, hunched posture, rough hair coat and partially closed eyelids (Supplementary Video 2). In contrast, while some of these signs were transiently present in control mice for the first few minutes after injection, they always subsided within a few minutes. Supporting vb− mice hyperalgesia (i.e. increased pain sensitivity), latency of tail withdrawal following focal application of radiant heat (tail-flick test46) was significantly shorter than in control mice (Fig. 4g; P < 0.005; n = 5 Ctrl and n = 4 vb− mice; KS test).

Taken together, these behavioral findings suggest that while vb− mice have a heightened sensitivity to pain (hyperalgesia), they do not appear to misperceive touch as a painful stimulus (allodynia, which is a clinically and experimentally distinct entity59), since they explore their environment to the same extent than control mice without detectable signs of pain.

While hyperalgesia in vb− mice may be due to the broader cortical receptive field of Po afferents as well as amplification of Po signals by strongly interconnected L4 neurons in S160, lack of allodynia suggests that there is little or no overlap between somesthetic and nociceptive pre-thalamic pathways following ablation of VB, in line with the largely preserved anatomical and functional segregation of ascending thalamic pathways discussed in Supplementary Note 5 above.

Supplementary Discussion The role of cell-intrinsic vs. input-dependent factors in specifying cortical phenotypes has been extensively explored in the past, using a variety of approaches including transplantation61,62, selective lesions of pre-cortical relays1,63, and genetic manipulation of pre-cortical neurons5,6,64,65 to distinguish the contribution of intrinsic versus local cues in neuronal differentiation.

In a seminal study61, Schlaggar and O’Leary have for instance shown that late embryonic rat visual cortex transplanted to neonatal somatosensory cortex develops barrel-like structures, providing a first indication that cortical neurons can be specified/respecified by incoming TC afferents. In a more recent study, ESC-derived neurons transplanted into the motor cortex remained seemingly impervious to their host area’s identity, since

W W W. N A T U R E . C O M / N A T U R E | 3

SUPPLEMENTARY INFORMATION RESEARCH


they expressed molecular markers and connectivity typical of visual cortex neurons62. Thus, the permissiveness of molecular differentiation programs to environmental cues may dependent on the precise developmental stage and molecular identity of the grafted neurons and their host tissue.

Changing the number of visual TC axons results in a corresponding change in the size of the primary visual cortex (V1)64, and loss of visual TC axons leads to loss of demarcation between V1 and neighboring visual areas5, together suggesting that TC input controls acquisition of areal identity from a genetically relatively uniform cortex66. Interestingly, in line with these findings, it has recently been proposed that rapid cortical expansion during evolution has “untethered” some cortical regions from constraints of molecular gradients and early activity cascades, resulting in emergence of sensory hierarchies.67

Synaptic release of glutamate from TC axons is required for cortical map formation6,68, but the postsynaptic cellular mechanisms controlling the arealization effects described above and whether they are accompanied by input-specific changes in the structure and function of intracortical circuits, remained unknown. Here, we identify three input-dependent genetic targets, Npas4, Zbtb20, and Sema3a which each function to polarize L4 dendrites towards incoming VB axons in S1, thereby critically controlling the microcolumnar assembly of cortical circuits. It will be interesting in future studies to investigate how these changes relate to the acquisition of S2L4-like excitatory/inhibitory synaptic response ratios in S1L4vb- and other area-specific circuit properties.

Supplementary References 47. Croquelois, A., Bronchti, G. & Welker, E. Cortical origin of

functional recovery in the somatosensory cortex of the adult mouse after thalamic lesion. Eur J Neurosci 21, 1798–1806 (2005).

48. Buch, T. et al. A Cre-inducible diphtheria toxin receptor mediates cell lineage ablation after toxin administration. Nat Meth 2, 419–426 (2005).

49. Matsui, A., et al. BTBD3 controls dendrite orientation toward active axons in mammalian neocortex. Science 342, 1114–1118 (2013).

50. Joshi, P. S., et al. Bhlhb5 regulates the postmitotic acquisition of area identities in layers II-V of the developing neocortex. Neuron 60, 258–272 (2008).

51. Kashani, A. H., et al. Calcium activation of the LMO4 transcription complex and its role in the patterning of thalamocortical connections. J Neurosci 26, 8398–8408 (2006).

52. Sato, H., et al. Thalamus-Derived Molecules Promote Survival and Dendritic Growth of Developing Cortical Neurons. J Neurosci 32,15388–15402 (2012).

53. Nielsen, J. V., Thomassen, M., Møllgård, K., Noraberg, J. & Jensen, N. A. Zbtb20 Defines a Hippocampal Neuronal Identity Through Direct Repression of Genes That Control Projection Neuron Development in the Isocortex. Cerebral Cortex doi:10.1093/cercor/bhs400 (2013).

54. Bloodgood, B. L., Sharma, N., Browne, H. A., Trepman, A. Z. & Greenberg, M. E. The activity-dependent transcription factor NPAS4 regulates domain-specific inhibition. Nature 503, 121–125 (2013).

55. Lin, Y., et al. Activity-dependent regulation of inhibitory synapse development by Npas4. Nature 455, 1198–1204 (2008).

56. Pouchelon, G., Frangeul, L., Rijli, F.M. & Jabaudon, D. Patterning of pre-thalamic somatosensory pathways. Eur J Neurosci 35, 1533–1539 (2012).

57. Veinante, P. & Deschenes, M. Single- and multi-whisker channels in the ascending projections from the principal trigeminal nucleus in the rat. J Neurosci 19, 5085–5095 (1999).

58. Lokmane, L., et al. Sensory Map Transfer to the Neocortex Relies on Pretarget Ordering of Thalamic Axons. Curr Biol, 23, 810–816 (2013).

59. Sandkühler, J. Models and mechanisms of hyperalgesia and allodynia. Physiol Rev 89, 707–758 (2009).

60. Beierlein, M., Fall, C.P., Rinzel, J. & Yuste, R. Thalamocortical bursts trigger recurrent activity in neocortical networks: layer 4 as a frequency-dependent gate. J Neurosci 22; 9885–9894 (2002).

61. Schlaggar, B.L. & O'Leary, D.D. Potential of visual cortex to develop an array of functional units unique to somatosensory cortex. Science 252,1556–1560 (1991).

62. Gaspard, N. et al. An intrinsic mechanism of corticogenesis from embryonic stem cells. Nature 455, 351–357 (2008).

63. Métin, C. & Frost, D.O. Visual responses of neurons in somatosensory cortex of hamsters with experimentally induced retinal projections to somatosensory thalamus. Proc Natl Acad Sci USA 86, 357–361 (1989).

64. Vue, T.Y., et al. Thalamic control of neocortical area formation in mice. J Neurosci 33, 8442–8453 (2013).

65. Nakagawa, Y., Johnson, J.E. & O'Leary, D.D. Graded and areal expression patterns of regulatory genes and cadherins in embryonic neocortex independent of thalamocortical input. J Neurosci 19, 10877–10885 (1999).

66. Pouchelon, G. & Jabaudon, D. Nurturing the cortexʼs thalamic nature. Curr Opin Neurol 27, 142–148 (2014).

67. Buckner, R.L. & Krienen, F.M. The evolution of distributed association networks in the human brain. Trends in Cognitive Sciences 17, 648–665 (2013).

68. Narboux-Nême, N, et al. Neurotransmitter release at the thalamocortical synapse instructs barrel formation but not axon patterning in the somatosensory cortex. J Neurosci 32, 6183–6196 (2012).

Modality-specific thalamocortical inputs instruct the identity of postsynaptic L4 neurons

Documents

Transcript of Modality-specific thalamocortical inputs instruct the identity of postsynaptic L4 neurons