Lateral geniculate neurons show robust ocular dominance ... - bioRxiv

bioRxiv preprint first posted online Jan. 3, 2017; doi: http://dx.doi.org/10.1101/097923. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. All rights reserved. No reuse allowed without permission.

Lateral geniculate neurons show robust ocular dominance plasticity Juliane Jäpel1, Mark Hübener1, Tobias Bonhoeffer1 and Tobias Rose1 1

Max Planck Institute of Neurobiology, Martinsried, Germany

Abstract Experience-dependent plasticity in the mature visual system is considered exclusively cortical. Using chronic two-photon Ca2+ imaging, we found evidence against this tenet: dLGN cells showed robust ocular dominance shifts after monocular deprivation. Most, but not all responses of dLGN cell boutons in binocular visual cortex were monocular during baseline. Following deprivation, however, deprived-eye dominated boutons became responsive to the non-deprived eye. Thus, plasticity of dLGN neurons contributes to cortical ocular dominance shifts.

Correspondence: [email protected] or [email protected] Max Planck Institute of Neurobiology Synapses – Circuits – Plasticity Am Klopferspitz 18 D-82152 Martinsried Phone: +49 89 8578 – 3684 Fax: +49 89 8578 – 2481

1


1

The change in ocular dominance (OD) after monocular deprivation (MD) is one of the most prominent

2

models of experience-dependent plasticity in the neocortex. Whereas MD is known to evoke an OD shift in

3

the binocular part of primary visual cortex (V1)1–4, robust OD changes in the dorsolateral geniculate nucleus

4

(dLGN) of the thalamus have not been reported so far2,5. This and similar findings led to the prevailing view

5

that after an early developmental phase of subcortical activity-dependent changes6, plasticity in the visual

6

system is exclusively cortical in the mature brain7,8. However, none of the recordings in dLGN to date have

7

been performed chronically with single-cell resolution. Hence, changes in the eye-specific responsiveness

8

of individual thalamic relay cells (TRCs) could have easily been missed. Furthermore, in contrast to the clas-

9

sical view of strict eye-specific segregation in dLGN9–11, some recent studies have reported that a substantial

10

fraction of rodent TRCs is binocular12,13, which could provide a substrate for competitive TRC plasticity. This

11

led us to re-address the questions of thalamic binocularity and experience-dependent OD plasticity, using

12

chronic two-photon Ca2+ imaging of TRCs projecting to binocular V1.

13 14

We conditionally expressed the genetically encoded Ca2+ indicator GCaMP6m14 in the dLGN of Scnn1a-Tg3-

15

Cre mice15 using adeno-associated virus16–19 and followed the visually evoked Ca2+ signals of individual TRC

16

axonal boutons in layer 1 (L1) of binocular V1 for up to 4 weeks before and after MD (Fig. 1a-d). Our chronic

17

bouton recordings provided a reliable and minimally-invasive longitudinal readout for the activity of TRCs

18

in dLGN, specifically (see Methods and Supplementary Fig. 1,2).

19 20

Recent electrophysiological studies have reported a large fraction of binocular neurons in rodent dLGN12,13.

21

It is, however, difficult to rule out that this could have been the result of imperfect single unit isolation.

22

Two-photon Ca2+ imaging circumvented this problem by allowing us to measure the OD of single boutons

23

along dLGN afferents in V1 (Fig. 1c-e). We indeed observed unambiguous responses to both eyes in a subset

24

of boutons (Fig. 1d). However, the binocularly responsive fraction (dLGN: 14%, Fig. 1e) was considerably

25

smaller than that reported in recent electrophysiological experiments13. When comparing the OD of dLGN

26

afferents with that of excitatory L2/3 cells in binocular V1, we found that, as expected, cortical neurons

27

were significantly more binocular than dLGN afferents (cortex: 42%, Fig. 1e, P < 10-3), whereas the ratio of

28

contralateral- over ipsilateral-eye population responses was comparable (Supplementary Fig. 3e,f, P =

29

0.36). Therefore, mouse V1 does not simply inherit binocularity from binocular geniculate cells, but, similar

30

to carnivorans and primates5, predominantly integrates eye-segregated dLGN input.

31 32

We compared the stimulus selectivity of binocular TRCs to their monocular counterparts and to excitatory

33

L2/3 cells in V1. We found, as expected18–21, that TRCs were far less orientation- and direction-selective

34

than cortical neurons (Fig. 1f,g; gOSI: P < 10-193; gDSI: P < 10-120). Similar to cortex, ipsilateral TRC boutons

35

tended to be slightly less sharply tuned than contralateral boutons (Fig. 1f,g). However, the stimulus selec-

36

tivity of binocular and monocular TRC boutons did not differ in any of the parameters we tested (Fig. 1f,g;

2


1

gOSI: P = 0.65; gDSI: P = 0.016, with ipsilateral and contralateral vs. binocular gDSI P = 0.11 and P = 1,

2

respectively), arguing against the possibility that binocular dLGN cells may represent a separate visual pro-

3

cessing channel in mice.

4 5

Does the mature dLGN show experience-dependent plasticity? To test this, we subjected adult mice to 6-8

6

days of contralateral eye MD (see Methods). Contrary to the prevailing view, we observed a clear (Fig. 2, P

7

< 10-8) and largely reversible (Supplementary Fig. 4) OD shift. Whereas most TRC boutons were monocular

8

during baseline, many acquired binocularity during MD (Fig. 2c). More than half of the responsive boutons

9

changed their ocular dominance index (ODI) significantly (>1 standard deviation of pre-MD baseline fluctu-

10

ations, Fig. 2d). In fact, the fraction of plastic boutons, the population shift magnitude, and the decrease in

11

the fraction of responsive boutons after MD (Fig. 2e) were comparable to the effects of MD in excitatory

12

L2/3 neurons in V14 (Supplementary Fig. 5a).

13 14

On the population level, the OD shift resulted from a net increase in ipsilateral-eye and a decrease in con-

15

tralateral-eye responsiveness (Supplementary Fig. 4e). On the level of individual geniculate afferents, bou-

16

tons that were exclusively responsive to the contralateral eye during baseline showed the most prominent

17

decrease in contralateral-eye evoked activity (Fig. 2f,g). This form of deprived-eye depression is similar to

18

our previous observation in excitatory L2/3 cells in V14 (Supplementary Fig. 5b,c). However, very different

19

from cortex, initially monocular contralateral boutons also showed the most prominent increase in ipsilat-

20

eral-eye responsiveness (compare Fig. 2f,g and Supplementary Fig. 5b,c).

21 22

We found that many TRC boutons, even if they appeared monocular during baseline, received robust input

23

from both eyes after MD (Fig. 2c). What are potential explanations for such a prominent OD change? One

24

– in our view less likely – explanation resides in the fact that the dLGN receives substantial modulatory

25

feedback from V122 and weaker input from superior colliculus23,24. A decrease in cortical deprived-eye re-

26

sponsiveness4 (Supplementary Fig. 5b,c) could therefore, in principle, propagate to the dLGN and contrib-

27

ute to a decrease in thalamic responsiveness to the deprived eye that mirrors the cell-specific changes in

28

L2/3 (Fig. 2f,g). We consider this less probable, however, because convergent modulatory cortical and sub-

29

cortical feedback would at the same time have to elicit highly target-specific changes in non-deprived-eye

30

responsiveness in order to lead to the specific strengthening of ipsilateral responses of initially contralater-

31

ally dominated TRCs (Fig. 2f,g). This is even more unlikely given the very different cortical changes in non-

32

deprived-eye response strength. Here, all excitatory L2/3 cells, except initially monocular contralateral cells,

33

increase their ipsilateral eye responsiveness after MD4 (Supplementary Fig. 5b,c). The, perhaps unex-

34

pected, but most parsimonious explanation is therefore, that the synapses of retinal ganglion cells (RGCs)

35

onto TRCs are the site of change. Recent connectomics studies report a dramatic mismatch between a large

36

structurally25–27 and much smaller functionally11,28 identified number of RGCs converging onto TRCs. We

3


1

therefore speculate that unsilencing of structurally present but functionally silent ipsilateral RGC input27,29

2

may be the reason for the robust OD plasticity we observe in the dLGN.

3 4

Regardless of the precise mechanism, our results clearly show that – contrary to current models – the eye-

5

specific responses of dLGN neurons are not rigid but undergo substantial functional OD plasticity. Given the

6

paramount importance of this paradigm for theories of experience-dependent plasticity and the ever more

7

widespread use of mice as model organisms, we conclude that one needs to exert caution when employing

8

purely cortical interpretations of OD plasticity.

9 10 11 12 13

Acknowledgements

14

We thank C. Huber, V. Staiger, F. Voss and H. Tultschin for technical assistance, J. Sigl-Glöckner for help with

15

our transfection protocol, and P. Goltstein for programming help. This work was supported by the Max

16

Planck Society, a Marie Curie Intra-European Fellowship to T.R., a Boehringer Ingelheim Ph.D. fellowship to

17

J.J., and the Deutsche Forschungsgemeinschaft (grant SFB870 to T.R. and T.B.). Data and analysis codes are

18

stored and curated at the Max Planck Computing and Data Facility Garching, Munich, Germany.

4


1

Figure legends

2 3

Figure 1. Long-term imaging and functional characterization of thalamic relay cell (TRC) boutons in L1 of

4

mouse V1

5

a) Schematic of imaging approach. b) Experimental timeline and two example image volumes illustrating

6

facilitated refinding of boutons over the extended time-course of the experiment (four slices at an image

7

plane depth increment of 3 μm, scale bar 10 μm, frame-averaged GCaMP6m fluorescence; see Methods

8

and Supplementary Fig. 1). c) Example of a color-coded response map of individual TRC boutons. Ocular

9

dominance (OD) is depicted as the pixel-wise peak fluorescence change in response to ipsi- and contrala-

10

teral eye preferred grating presentation (scale bar 10 μm). Red hues indicate ipsilateral dominance (ocular

11

dominance index, ODI = (∆F/Fcontra − ∆F/Fipsi) / (∆F/Fcontra + ∆F/Fipsi); ODI < 0), and blue hues indicate contra-

12

lateral dominance (ODI > 0). d) Ca2+ signals in a binocular (top) and two monocular example boutons in

13

response to ipsi- and contralateral eye drifting grating stimulation (scale bars: ΔF/F0 = 200 %, 10 s). e) Frac-

14

tion of responsive dLGN boutons and excitatory L2/3 cell bodies in binocular V14 showing purely ipsilateral,

15

purely contralateral or binocular significant responses (dLGN, n = 725 boutons, 9 mice; V1, n = 2.095 cells,

16

14 mice; P < 10-3, χ² test). f, g) Violin plots of gOSI (f) and gDSI (g) of ipsi-, contralateral and binocular dLGN

17

boutons and L2/3 cell bodies, respectively. Horizontal lines indicate medians, shaded areas the mirrored

18

probability density estimates (dLGN, n = 725 boutons, 9 mice; V1, n = 2.095 cells, 14 mice; all gOSI dLGN vs.

19

V1, P < 10-193, all gDSI dLGN vs. V1, P < 10-120, Mann-Whitney U-test; gOSI dLGN, P = 0.65; gDSI dLGN, P =

20

0.016, Kruskal-Wallis test, ipsi- or contralateral vs. binocular P = 0.11 and P = 1, Bonferroni corrected; gOSI

21

V1, P < 10-6; gDSI V1, P = 0.066, Kruskal-Wallis test, ipsi- vs. contralateral or binocular gOSI P < 10-5 and P
8 * σΔF/F0, Baseline) was observed in at least 50% of the trials

3

of a single simulation condition for a specific eye. The published dataset of excitatory L2/3 cells6 was

4

analyzed as described previously6, with the exception that now the same eye-specific responsiveness

5

criterion as for thalamocortical boutons was used. This led to minor differences in cell selection in

6

comparison to our previous analysis (Supplementary Fig. 5, compare to Fig. 2d in Rose et al. 20166).

7 8

Data analysis

9

Ocular dominance was determined by calculating the ODI from the mean ΔF/F0 response to the preferred

10

eye-specific grating direction: ∆𝐹 𝑐𝑜𝑛𝑡𝑟𝑎𝑝𝑟𝑒𝑓_𝑑𝑖𝑟 − 𝐹 𝑂𝐷𝐼 = 0 ∆𝐹 𝐹0 𝑐𝑜𝑛𝑡𝑟𝑎𝑝𝑟𝑒𝑓_𝑑𝑖𝑟 +

11

∆𝐹 𝑖𝑝𝑠𝑖𝑝𝑟𝑒𝑓_𝑑𝑖𝑟 𝐹0 ∆𝐹 𝐹0 𝑖𝑝𝑠𝑖𝑝𝑟𝑒𝑓_𝑑𝑖𝑟

12

Contralateral and ipsilateral dominance are indicated by an ODI of 1 or -1, respectively. For chronic single-

13

bouton changes in ODI, only boutons that were responsive throughout baseline and after MD (Fig. 2c), and,

14

if applicable, during recovery (Supplementary Fig. 4f-h) were considered. For the quantification of changes

15

in eye-specific response amplitude relative to the mean baseline ODI the 3 baseline sessions had to be

16

responsive (Fig. 2f,g; Supplementary Fig. 5c,d). Pixel-based color-coded ODI maps were similarly generated

17

by obtaining ODI values from the maximum stimulus evoked fluorescence change of individual pixels. Hue

18

was set by the ODI (red: ipsilateral dominance, blue: contralateral dominance) and lightness coded the

19

mean of the summed ipsilateral and contralateral response amplitude.

20 21

The Contra/ipsi ratio was calculated as the ratio of the mean fluorescence response to ipsilateral or

22

contralateral eye stimulation, respectively, over all responsive boutons or layer 2/3 cell bodies of one

23

animal (Supplementary Fig. 3e,f).

24 25

A global orientation selectivity index (gOSI) was calculated as 1 – circular variance (circ. var.)13: 𝑔𝑂𝑆𝐼 = 1 − 𝑐𝑖𝑟𝑐. 𝑣𝑎𝑟. = |

26

∑ 𝑅(𝜃𝑘 )𝑒 2𝑖𝜃𝑘 | ∑ 𝑅(𝜃𝑘 )

27

with R(θk) as the mean ΔF/F0 response to the orientation angle θk. Perfect orientation selectivity is indicated

28

as gOSI = 1. Similarly, direction tuning was assessed by calculating the global direction selectivity index

29

(gDSI) 13:

30

𝑔𝐷𝑆𝐼 = 1 − 𝑑𝑖𝑟. 𝑐𝑖𝑟𝑐. 𝑣𝑎𝑟. = |

∑ 𝑅(𝜃𝑘 )𝑒 𝑖𝜃𝑘 | ∑ 𝑅(𝜃𝑘 )

31 32

For analysis of receptive fields (RFs), traces were sorted according to stimulus position and type (black or

33

white, representing ON and OFF subfields). Response latency was defined as the first imaging frame during

6


1

a set time window in which there was a significant increase (ANOVA, p < 0.05) in fluorescence. If the ROI

2

was determined as responsive for both the black (OFF) and white (ON) patches with different latencies,

3

only the RF with the shorter latency was considered for further analysis. A thresholded RF subfield was

4

derived by interpolating the raw RF at 1° resolution and only taking the region with the strongest average

5

response. This subfield was used to compute RF area.

6 7

Statistics

8

Data are reported as median ± interquartile range (IQR) or as mean ± standard error of the mean (SEM) or

9

standard deviation (ST) as indicated in individual figures. Paired parametric (t-test) or paired and unpaired

10

non-parametric (Wilcoxon signed-rank test, Mann-Whitney U test, Kruskal-Wallis test with Bonferroni

11

posttest) statistics were used for comparison. For cumulative distributions, the Kolmogorov-Smirnov test

12

was applied, while distributions were compared using χ² test. Asterisks indicate significance values as

13

follows: * p < 0.05, ** p < 0.01 and *** p < 0.001.

7


1

Supplementary Figure legends

2 3 4

Supplementary Figure 1. dLGN virus injections and matching of regions of interest (ROIs) across sessions

5

a) AAV coding for GCaMP6m was injected into the dLGN of Scnn1A-Tg3-Cre-mice expressing Cre-

6

recombinase in cortical layer 4 and thalamus. The representative coronal section shows GCaMP6m-labeled

7

neurons in the dLGN (outlined in right panel, scale bar: left 500 μm, right: 200 μm). Note absence of

8

transfected cells in the injection path and negligible dorsomedial spread into the lateral posterior nucleus

9

(LP). b) Example image volumes illustrating matching of bouton ROIs across sessions. Changes in tissue

10

morphology and positioning inaccuracies can result in individual boutons being optimally sectioned by

11

different image planes at different time points (arrows of the same color indicate ROIs of the same group,

12

image plane depth increment: 3 μm, first frame of volume 20 µm below pial surface; scale bar: 10 μm).

13 14

Supplementary Figure 2. Spatial receptive field properties of thalamic relay cell (TRC) boutons in L1 of V1

15

a) Example Ca2+ traces (average of 10 trials) of a single bouton responding to pseudorandomized sparse

16

noise stimulation (white, left, and black, right, patches on an isoluminant gray background) of the

17

contralateral eye, ordered according to stimulus position (scale bars: ΔF/F0 = 50%, 1 s). b) Receptive field

18

of the corresponding bouton in (a). Bottom plot shows outline of the ON and OFF sub-fields (light: ON,

19

dark: OFF). c) Violin plots of spatial receptive field size of dLGN boutons and excitatory layer 2/3 cell bodies

20

in V1 to ipsilateral (red) or contralateral (blue) eye stimulation, respectively. Horizontal lines indicate

21

medians, shaded areas the mirrored probability density estimates (dLGN, n = 1198 receptive fields, 9 mice;

22

V1, n = 123 receptive fields, 4 mice; combined median receptive field area dLGN: 195 ± 88 deg², V1: 208 ±

23

99 deg², P = 0.18, Wilcoxon rank-sum test).

24 25

Supplementary Figure 3. Population orientation- / direction-selectivity and eye-specific responsiveness of

26

TRC boutons in L1 of V1

27

a) Ca2+ signals of three orientation-selective example boutons in response to ipsi- and contralateral eye

28

drifting grating stimulation (scale bars: ΔF/F0 = 200%, 10 s). b) Polar plots displaying response amplitude

29

(in ΔF/F0) of the boutons shown in (a). Numbers indicate global orientation selectivity indices (gOSI = 1 –

30

circular variance). c, d) Fraction of orientation- (c, gOSI > 0.33) and direction-selective (d, gDSI > 0.33) ipsi-

31

, contralateral and binocular dLGN boutons and V1 layer 2/3 cell bodies, respectively (dLGN, n = 9 mice;

32

V1, n = 14 mice; all gOSI dLGN vs. V1, P < 10-10, all gDSI dLGN vs. V1, P < 10-10, Mann-Whitney U-test; gOSI

33

dLGN, P = 0.62, gDSI dLGN, P = 0.99, gOSI V1, P = 0.20, gDSI V1, P = 0.14, Kruskal-Wallis test). e) Normalized

34

contra- (blue) and ipsilalateral (red) population response of dLGN boutons (upper traces) and excitatory

35

L2/3 cell bodies in V1 (lower traces; preferred drifting grating stimulation, normalized to peak contralateral

36

response, dashed lines; dLGN, contra/ipsi = 2.3, n = 725 boutons; V1, contra/ipsi = 2.0, n = 2095 cells).

8


1

Arrowhead indicates the pronounced dLGN on-response to pseudorandom eye-shutter switches, which is

2

not present in cortex. f) Population contra/ipsi ratio in dLGN and V1 (dLGN, n = 9 mice, V1, n = 14 mice, P

3

= 0.36. Wilcoxon rank-sum test).

4 5

Supplementary Figure 4. OD plasticity and recovery from OD plasticity in TRC boutons

6

a) Schematic of stimulus presentation for intrinsic optical imaging. Visual stimuli were presented at two

7

different spatial locations on a screen in front of the mouse. b) Color-coded maps of cortical responses to

8

stimulation of the ipsilateral (upper panel) or contralateral (lower panel) eye (scale bar: 0.5 mm), overlaid

9

on an image of the cortical surface blood vessel pattern. c) ODI values based on intrinsic optical imaging of

10

cortical responses for normal adult mice (mean ± SEM: 0.26 ± 0.03, n =9), after 6-8 days of MD (0.04 ± 0.04,

11

n = 9), and after 1-2 weeks of recovery (0.15 ± 0.05, n = 6 animals, P < 10-4, Kruskal-Wallis test with

12

Bonferroni’s post hoc test for multiple comparisons, baseline vs. MD P = 0.0006, baseline vs. recovery P =

13

0.45, MD vs. recovery P = 0.41). d) Example time course of dLGN single-bouton ODI (gray lines) from one

14

animal over baseline, contralateral eye MD and recovery (black line: mean ODI). e) Cumulative eye-specific

15

response amplitude before and after MD (n = 456 dLGN boutons, responsive during baseline and post-MD,

16

ipsilateral eye responses baseline vs. post-MD P < 10-5, contralateral eye responses baseline vs. post-MD P

17

< 10-6, two-sample Kolmogorov-Smirnov tests). f) ODI of boutons (n = 168 boutons from 5 mice, responsive

18

during baseline, MD and recovery) after MD and at least one week of binocular vision (P < 10-16, baseline

19

vs. MD P > 10-6, baseline vs. recovery P = 0.002, MD vs. recovery P< 10-10, Friedman’s test with Bonferroni’s

20

post hoc test for multiple comparisons). g) Response amplitude to ipsilateral eye stimulation after MD and

21

at least one week of binocular vision (P < 10-13, baseline vs. MD: P = 0.01, baseline vs. recovery P < 10-6, MD

22

vs. recovery P < 10-13, Friedman’s test with Bonferroni post hoc for multiple comparisons). h) Response

23

amplitude to contralateral eye stimulation after MD and at least one week of binocular vision (P < 10-6,

24

baseline vs. MD P < 10-6, baseline vs. recovery P = 0.055, MD vs. recovery P = 0.035, Friedman’s test with

25

Bonferroni post hoc for multiple comparisons).

26 27

Supplementary Figure 5. Cell-specificity of OD plasticity in excitatory layer 2/3 cells in V1

28

a) ODI distribution of the same excitatory L2/3 cells in V1 during baseline (mean of three pre-MD sessions,

29

0.28 ± 0.47, SD) and after 5-8 days of MD (0.07 ± 0.67, SD, n = 430 cells in 10 mice, P < 10-11, Wilcoxon

30

signed-rank test). Lines connect individual, continuously responsive cells; line color indicates shift

31

significance in units of standard deviations over pre-MD baseline fluctuations. Colored ODI histogram bins

32

indicate class definitions for contralateral (blue), binocular (white) and ipsilateral (red) boutons (data

33

reanalyzed from Rose et al. 20166). b) Change in eye-specific response amplitude after MD for individual

34

cells (n = 565 cells responsive in all pre-MD sessions, 10 animals). Cells are color-coded by their mean pre-

35

MD ODI. c) Quantification of b), grouped into pre-MD ODI sextiles comprising similar numbers of cells

9


1

(94-95 cells per group, paired t-tests; monocular ODI classes indicated by red and blue shading as

2

defined in (a) and Fig. 2c,g).

10


1

Supplementary References

2 3

1.

Madisen, L. et al. Nat. Neurosci. 13, 133–140 (2010).

4

2.

Roth, M.M. et al. Nat. Neurosci. 19, 299–307 (2016).

5

3.

Kondo, S. & Ohki, K. Nat. Neurosci. 19, 316–319 (2015).

6

4.

Sun, W., Tan, Z., Mensh, B.D. & Ji, N. Nat. Neurosci. 19, 308–15 (2016).

7

5.

Durand, S. et al. J. Neurosci. 36, 12144–12156 (2016).

8

6.

Rose, T., Jaepel, J., Hübener, M. & Bonhoeffer, T. Science 352, 1319–22 (2016).

9

7.

Hofer, S.B., Mrsic-Flogel, T.D., Bonhoeffer, T. & Hübener, M. Nat. Neurosci. 9, 127–132 (2006).

10

8.

Greifzu, F. et al. Proc. Natl. Acad. Sci. 111, 1150–1155 (2014).

11

9.

Matthies, U., Balog, J. & Lehmann, K. J. Neurosci. 33, 11774–8 (2013).

12

10.

Pologruto, T.A., Sabatini, B.L. & Svoboda, K. Biomed. Eng. Online 2, 13 (2003).

13

11.

Pelli, D.G. Spat. Vis. 10, 437–442 (1997).

14

12.

Brainard, D.H. Spat. Vis. 10, 433–436 (1997).

15

13.

Mazurek, M., Kager, M. & Van Hooser, S.D. Front. Neural Circuits 8, 92 (2014).

16

11

a

b

session 1 (d0)

session 4 (d15)

-20 μm

-23 μm

-20 μm

-23 μm

-26 μm

-29 μm

-26 μm

-29 μm

Supplementary Figure 1

b

c

n. s.

600 500

ON

OFF

40 20 0 −20

Area (deg2)

a

400 300 200 100

−40 −20 0 20

0


dLGN

V1

ipsi

b

contra

ROI 1, ipsi 60

30

1

200 100

n. s.

80 60 40 20

ipsi contra

0

binocular

dLGN

V1

Percentage direction−selective boutons (gDSI > 0.33)

n. s.

400

ROI 3, ipsi

0.63 270

240

200 100

0.81 270

n. s.

90 400 200 100

210 300

e

90 400

30

0.77

240

270

210 300

0.62

240

270

n. s. 10

40 20

8 6 4 2

dLGN

V1


V1

0

180

330

f

dLGN

150

200 100

n. s.

60

120

300

180 0

330

240

60 150

80

0

ROI 3, contra 120

300

180 0 210

300

30

150

*** 100

60

120

300

180 0

d

*** 100

30

90

210 330 300

c

60 150

330

3

Percentage orientation−selective boutons (gOSI > 0.33)

400

ROI 2, contra 120

300

0

2

90

Contra/ipsi ratio

a

dLGN

V1

c

d

n. s. ***

0.3

1

0.2

*

800

600

400

200

0

ry ve co

ry ve

ry ve co


n. s.

D

0

D

−1

re

800 1000

-M

400 600 ∆F/F0

st

200

po

ipsi − post−MD

0

pr ea e- n M D

0

200

***

15

re

contra − post−MD

10

-M

−0.5

ipsi − mean pre−MD

co

contra − mean pre−MD

400

D

0.25

0

1000

600

-M

0.5

h

***

800

st

ODI

0.5

***

5 days

st

Response amplitude (∆F/F0), ipsi eye

0.75

*

0

po

1000

−5

m e e- an M D

g

***

−1

recovery

pr

**

***

m

Cumulative fraction of boutons

***

post-MD

po

1

***

pre-MD

m e e- an M D

f 1

0

Response amplitude (∆F/F0), contra eye

-0.1 contra eye

e

recovery

−0.5

pr

M P

0.1 0

A L

post-MD

0.5 ODI

20°

ipsi eye

ODI

40°

pre-MD mean pre-MD

re

b

a

***

b

post-MD

0 1

fold σbaseline

ODI

0.5 0

−0.5

−1

5 100 cells

c

50

600 400 200 0 -200 -400 -400

-200

0

200

400

600

Response amplitude change deprived eye [% ∆R/R0]


Response amplitude change [% ∆R/R0]

mean pre-MD

mean pre-MD ODI c i -1 1

1

Response amplitude change non−deprived eye [% ∆R/R0 ]

a

**

**

** **

0

***

-50

-100 -1

**

contra ipsi 0.5

0

*** -0.5

Mean pre−MD ODI

1