Round 1

Reviewer 1 Report

This study reports that inactivation of cat pulvinar influences the oscillatory nature of visually-driven responses in visual cortical areas 17 and 21a, as well as the coupling between these areas. It is an entirely plausible hypothesis for cortical coupling to depend on thalamic links, and this is a worthwhile experiment. My fundamental criticism of this work is that much of the text inadequately explains the experiments and data, and the effects shown are generally quite small. Frankly it would be just as interesting if the core result were that inactivation of pulvinar had NO effect on cortical communication. But if the authors want to convince the reader otherwise they need to do a better job.

Even in the abstract, the motivation for the study is somewhat unclear. The abstract states “Results indicate that cortical oscillatory activity was enhanced during pulvinar inactivation.” But a later sentences states “Granger causality showed that the pulvinar transferred oscillatory information between areas 17 and 21a in gamma and alpha bands for the feedforward and feedback processing, respectively.” This seems contradictory to state both that the pulvinar transfers oscillatory information, but also that pulvinar inactivation INCREASES oscillations. A clearer and more specific statement of the hypothesis guiding the study would be useful – not simply that the pulvinar Is somehow involved.

Methods: A number of methodological details need to be clarified

a) How many cats were studied?

b) Visual stimuli

line 103, it makes no sense to describe the screen as isoluminant. The visual stimuli were isoluminant to the screen? Certainly not. Did stimuli really subtend 116x150 deg visual angle? That is a very large display (at 57 cm, that would be 116x150 cm. Were they using a projection display?

c) Response quantification

2secs seems a somewhat long stimulus duration. Why was this duration chosen? Were responses measured and/or quantified during the entire 2 sec epoch, or were initial response transients ever discounted?

Line 111: Unclear if this means that each stimulus was separated from the next by 1 sec of blank or 0.5 sec. Authors should clarify.

Had responses always fallen to baseline after each stimulus before next one was presented?

Line 125 MUA response definition – why is area under the curve appropriate? Some justification would be helpful so the reader can understand this measure. Is it standard?

d) Line 148 Was Granger Causality measure calculated over entire stimulus epoch?

e) Inactivation

How did the authors assess that there were no indirect effects on LGN (and thus V1) responses? The authors never stated why they focus on LPl rather than any other portion of the LP/pulvinar complex; they should do so.

They never stated or showed the spread of inactivation within the LP pulvinar complex. Their stimuli seem to have subtended 116x150 deg visual angle (see above); in any case they extended across much of the visual field. Was all of the visually-activated pulvinar also chemically inactivated? I presume that much of the visually-activated pulvinar was NOT inactivated by their GABA injections. How might this influence their results?

Results – Much of my response to this study is that the results reported are of small magnitude effects, even when they were statistically significant. So I question how meaningful these results are. It was also very frustrating and confusing that the color code for response conditions was inconsistent. Figure 1 legend for example states injection was red trace, control was blue, and recovery black. But the inset on Fig1 itself shows control red, injection blue, and recovery black. This color code needs to be checked for the entire manuscript.

In all figures in this manuscript the authors should state what stimulus was being used. Was it the one that evoked the maximal response at that site? (or some average across all stimulus conditions) If not, how was stimulus condition used to quantify responses? And was this consistent across all analyses described in this manuscript?

 

Fig1

a) Fig 1 shows only small effects at high frequencies. Is this significant or meaningful? And why is there a DECREASE at low frequencies (10-20 Hz) that is not discussed?

b) 1.5 sec are shown, but stimulus duration was reported as 2 sec?

c) Rows 1 and 2 do not convincingly show responses before and after injection being similar. If responses did not recover to control condition, all analyses are suspect because of nonstationarity of responses. This needs to be clarified.

d) Fig 1F legend is unclear. How can the same electrode be in a different area? Do the authors mean a different electrode recording with responses all measured from that same electrode? Are the responses in panels G and H from the same epoch as D,E,F?

Fig2: I note in all panels a prominent change in power during the onset portion of the response? Does this affect authors’ interpretation of the results?

In Figure 3 the correlations shown are quite weak. But this pools recording sites in all layers in each area, which could mask potential effects. The authors do later in the paper separate data by later, but here they could also show them separately by layer (for example by using different symbols in the scatter plots for data from different layers, and calculating separate correlations).

These data have another problem which the authors themselves describe on line 290; the responses did not in fact recover to control conditions after GABA injection. The authors use a new criterion of selecting responses in which responses during injection period were larger than during the recovery period.

But why is this appropriate? Why not simply select responses that reversed fully? (i.e. they analyzed responses that were NOT selected to be the SAME during recovery and control conditions. So the authors were selecting for response epochs in which something else was going on. This is a big problem.

Fig 4 These selected responses are analyzed in Figure 4. Again they are pooling layers (which might mask effects); I see no effect, at best a tiny effect. (Again, they need to use a consistent color scheme)

Similarly in Figure 5 they show small magnitude effects, again presumably from this sample chosen with nonstationary responses as above.

What does it mean that so many responses (power changes) were not different from baseline even in control condition? This suggests weak neuronal responses; could this for example be due to the extended grating stimuli used, which might not have been effective at driving cells with suppressive surrounds known to be prominent in visual cortex?

Fig6 Granger Causality: Fig5 shows that in area21a alpha and gamma power did not change much in layer 5 (which provides much of the feedback signal), so why might one expect changes here in Fig6? And the authors should justify focusing on layer 5 only since cortical feedback can arise from layer 6 as well.

Were responses analyzed here the same selected (possibly nonstationary) ones shown in Figs 4 & 5?  

Data do not seem to match authors’ description in the text. (starting with line 336) I do not see the 30hz peak in every condition, so the authors should be clearer.

The authors should remind the reader what increased values of GC index mean. This is not adequately explained.

Line 349 not influences but indices. I do not understand what the authors are trying to say here.

Why offset of shadowed area from trace in lower left panel?

The red and blue traces were not identified as control and injection conditions in either the text or figure legend.

Discussion is overlong, and should be condensed considerably.

 

The language throughout needs improvement and careful proofreading.

Abstract 2^nd sentence is unclear. (Gamma waves influence cortical processing or reflect differences in processing? And do they really mean that the alpha oscillations themselves travel between areas?)

Abstract line 18 “alpha band responses were highly represented” is unclear

Line 20 “gamma oscillations…had more significant responses” (how can oscillations have responses?)

Line 23 “cortical oscillatory communication…” “cortical” is repetitive/unnecessary

Line 38 “contrast modulation” should be “contrast response” or “contrast sensitivity”

Line 40 “feedforwardly” is not correct English

Line 41 how can pathways represent oscillatory waves? Incorrect sentence

Line 69 “Cortical oscillatory responses that were layer-dependent” is not a sentence.

Line 72 “rose” should “arose”                 

Line 73 I do not understand the last sentence of Introduction.

Line 86 “ringer” should “Ringer’s”

Line 95 “eyes” should be “ the eyes’ ”

Line 128 “in LFPs” should be “on LFP data”

Line 129 “Power spectrum S was obtained BY multiplying…”   word “by” is missing

Line 143 (7.5 and 12.5 Hz) should be (7.5 to 12.5 Hz)

Lines 274-281 text was repeated “Consequently…”

Line 331: “…layers in area 17 has…” “has” should be “have”

Line 331: measurements were not analyzed, responses were

Line 332 unclear

Line 339 values not responses

Line 560 reference volume missing

Several references were all in italics (for ex refs 52-58)

Author Response

We thank reviewer 1 for his/her positive assessment of the manuscript. We have made all the changes suggested. Details are given below.

1. How many cats were studied?

We used two cats. We added this information in the text.

2. Visual stimuli

3. line 103, it makes no sense to describe the screen as isoluminant. The visual stimuli were isoluminant to the screen? Certainly not. Did stimuli really subtend 116x150 deg visual angle? That is a very large display (at 57 cm, that would be 116x150 cm. Were they using a projection display?

We were not clear here as pointed out by the reviewer. This was corrected.

4. 2secs seems a somewhat long stimulus duration. Why was this duration chosen? Were responses measured and/or quantified during the entire 2 sec epoch, or were initial response transients ever discounted?

The visual stimulation lasted two seconds to better investigate the dynamics of responses. Moreover, this duration ensured that the receptive field of the neuron population was repeatedly stimulated by the sinusoidal cycles of the drifting grating. Only the blank period before the visual stimulation was discounted.

4. Line 111: Unclear if this means that each stimulus was separated from the next by 1 sec of blank or 0.5 sec. Authors should clarify.

It is now better explained.

6. Had responses always fallen to baseline after each stimulus before next one was presented?

Yes. Figure 1 was reorganized. Averages of responses for MUAe and spectral powers are now shown. Responses show that the MUAe levels return to baseline before the stimulus was presented.

7. line 125 MUA response definition – why is area under the curve appropriate? Some justification would be helpful so the reader can understand this measure. Is it standard?

This method has been proposed by Xing et al., 2009, and provides an instantaneous signal of the spike activity of the neuron population near the tip of the electrode. Full-wave rectifying and low-pass-filtering the signal generating the MUA. This quantification provides a more direct measure than the method of recording multiple action potentials from different neurons by imposing an arbitrary threshold. The measure has been largely used by Roelfselma’s team (Self et al., 2012; Self et al., 2013; Pooresmaeili et al., 2014, van Kerkoerle et al., 2014) and it represents the firing rate activity in a dynamic way. Self et al., 2012 used the envelope of the signal (MUAe) to report the instantaneous firing rate; we have used the area under the curve to measure neuronal responses in a similar way than the oscillatory activity of alpha and gamma waves. However, the envelope describes the tangential integration of a mathematical continuous function, which is the integral of the continuous function with integration steps going asymptotically to infinity. For us, the area under the curve reflected the integral of the MUA. We have modified the text to clarify this concept.

8. Line 148 Was Granger Causality measure calculated over entire stimulus epoch?

In the current work, windows of 500 ms were used to calculate Granger causality measures. The length of such windows minimized possible non-stationary variations of neuronal responses throughout the visual stimulation period (Cohen, 2014). We have modified the text for clarification.

9. Inactivation. How did the authors assess that there were no indirect effects on LGN (and thus V1) responses? The authors never stated why they focus on LPl rather than any other portion of the LP/pulvinar complex; they should do so.

In cats, the pulvinar is described as a group of three nuclei denominated the lateral posterior-pulvinar complex (LP-pulvinar, (Hutchins and Updyke 1989)). The lateral part of the LP (LPl) is the striato-recipient zone of the LP while the medial part (LPm) is the tectorecipient zone. The LPl was the subdivision target by GABA injections because it is the only subdivision of the cat pulvinar that is connected reciprocally to both areas 17 and 21a (Berson and Graybiel 1983; Casanova 1993). The injections were restricted to this region and did not spread to the LGN (see below).

10. They never stated or showed the spread of inactivation within the LP pulvinar complex. Their stimuli seem to have subtended 116x150 deg visual angle (see above); in any case they extended across much of the visual field. Was all of the visually-activated pulvinar also chemically inactivated? I presume that much of the visually-activated pulvinar was NOT inactivated by their GABA injections. How might this influence their results?

The validation of GABA injection in the LP nucleus was performed by a series of coronal sections of the cat thalamus stained for Acetylcholinesterase showing the spread of the Chicago Sky Blue dye in the lateral subdivision of LP. Moreover, recordings were simultaneously in that region to quantify whether the inactivation was successful. After cessation of the injection, a recovery of neuronal activity was observed. Besides, we estimated the LP scotoma by comparing the extent of the Chicago Sky Blue staining with the retinotopic maps described by Hutchins and Updyke (1989) as we did in de Souza et al., (2019). A new figure has been added to the supplementary material.

 

11. Results – Much of my response to this study is that the results reported are of small magnitude effects, even when they were statistically significant. So I question how meaningful these results are. It was also very frustrating and confusing that the color code for response conditions was inconsistent. Figure 1 legend for example states injection was red trace, control was blue, and recovery black. But the inset on Fig1 itself shows control red, injection blue, and recovery black. This color code needs to be checked for the entire manuscript.

Corrected. We used the same colours consistently throughout the article.

12. Fig 1 shows only small effects at high frequencies. Is this significant or meaningful? And why is there a DECREASE at low frequencies (10-20 Hz) that is not discussed?

Following the suggestion of both reviewers, we have reorganized Figure 1 for clarity. The figure now shows the averages for the MUAe and the spectral powers of the LFP. In both cases, the distribution of averages was calculated by bootstrapping to distinguish whether average differences are significant. The spectral power of the LFP was also plotted on a logarithmic scale to better observe the differences. The power spectral density follows a 1/f distribution, it is inversely proportional to the frequency of the signal (Cohen, 2014). So high frequencies are less represented than low frequencies.

13. 1.5 sec are shown, but stimulus duration was reported as 2 sec?

We have included the latest periods in the figures.

14. Rows 1 and 2 do not convincingly show responses before and after injection being similar.

We provide new graphics that will be more convincing. Also, we have added a new index that shows percentage differences in conditions during and after the inactivation of the pulvinar.

15. If responses did not recover to control condition, all analyses are suspect because of nonstationarity of responses. This needs to be clarified.

Yes, this is true. The oscillatory changes that occurred throughout the experiment may have violated the stationarity profile of the responses. These non-stationary responses change the amplitude in the Fourier spectrum, representing more complicated frequency structures, and requiring more energy at higher frequencies in order to represent the time series in the frequency domain. The use of wavelet convolution avoids such a problem because this method assumes that the data is stationary within a short period (a few hundred milliseconds for the data) (Cohen, 2014). The second benefit of wavelet convolution is the ability to detect changes in frequency structure over time. Thus, the Morlet wavelet convolution provides the means to detect and analyze non-stationary signals (Van Drongelen, 2007). This is now explicit in the revised version.

16. Fig 1F legend is unclear. How can the same electrode be in a different area? Do the authors mean a different electrode recording with responses all measured from that same electrode?

The records were made simultaneously in areas 17 and 21a. Each area had an electrode inserted and recorded independently. We have modified the legend of figure 1 to clarify this point.

17. Are the responses in panels G and H from the same epoch as D,E,F?

Yes, they were. We rewrote the legend of figure 1 to clarify the origin of the electrophysiological signals.

 

18. Fig2: I note in all panels a prominent change in power during the onset portion of the response? Does this affect authors’ interpretation of the results?

We apologize for not having been clear enough in our manuscript. One explanation for this transient response is that the inactivation of LPl generated a stronger response in the feedforward pathway than the feedback pathway along the visual cortex. So, the feedforward activity could exclude the oscillatory action of the feedback pathway (Spaak, 2012). Another possibility for the transient alpha-wave response is that the duration of the stimulus was long enough to produce inhibition of the oscillatory activity by surround suppression (Vanni and Casanova, 2013). This suppression results in the alpha band's oscillatory activity not being as affected. It should be noted, moreover, that the animals are under anesthesia, and this can vary the performance of alpha and gamma oscillatory activities considerably (Hudetz et al., 2011; Rosier et al., 1993). We had addressed your comment in the discussion.

19. In Figure 3 the correlations shown are quite weak. But this pools recording sites in all layers in each area, which could mask potential effects. The authors do later in the paper separate data by later, but here they could also show them separately by layer (for example by using different symbols in the scatter plots for data from different layers, and calculating separate correlations).

Correlations between the MUAe and the AUC of alpha- and gamma waves were calculated for each layer of areas 17 and 21a. Since the difference between the injection and control periods was calculated, a positive correlation means an increase in oscillatory activity, in conjunction with the MUAe. In general, positive correlations were mild, and they were more pronounced for gamma-band oscillations. In area 17, layer VI had a significant trend (r=0.27, p<0.05), and layer IV approached the borderline of significance (r=0.23, p=0.06). Broader trends were observed in area 21a (All r-values>0.4). Here, all layers, except for the layer I, had statistically significant correlations (All p-values<0.01). For area 21a, the increase between MUAe and gamma waves follows the trend observed for the whole neuronal populations across cortical layers. This analysis was added to the supplementary material of the article.

 

20. These data have another problem which the authors themselves describe on line 290; the responses did not in fact recover to control conditions after GABA injection. The authors use a new criterion of selecting responses in which responses during injection period were larger than during the recovery period.

The selection made to analyze the relationship between LFP power, and multi-unit activity was based on the article by van Kerkoerle, et al. (2014). These authors described that the oscillations of the alpha and gamma band in the V1 area were negatively and positively correlated with the MUAe, respectively. These oscillatory cortical patterns were found in monkeys performing an attentional task. Since anesthetized animals were used in the current study, one way to reproduce similar conditions was to use MUAe as a measure of oscillatory response selection. The changes in MUAe produced by the inactivation of the thalamus would mimic the attentional processes. MUAe was used then to select the oscillations that followed a similar trend to those observed in this previous work.

 

21. But why is this appropriate? Why not simply select responses that reversed fully? (i.e. they analyzed responses that were NOT selected to be the SAME during recovery and control conditions. So the authors were selecting for response epochs in which something else was going on. This is a big problem.

We are sorry that the selection process was not well explained. Two selection criteria were used for the same visual evoked response (VER). The first, as explained above, used the MUAe to classify neuronal states similar to those observed during attentional tasks. Only the VERs that passed the first selection were considered. The second selection consisted of taking the oscillatory activity that goes to or reversed completely to control values (Rec-Inj>0). This second selection filtered oscillatory data that enhanced after inactivation, and so, to avoid a non-stationary process. Only the VERs that followed both criteria were analyzed (which is why it was possible to perform paired statistics). A similar selection process was carried out in our previous article for the classification of spike activity (de Souza et al, 2019).

We have not used VERs that reversed completely (MUAe or LFP, Rec-Ctr>0) as the only selection criteria because these responses did not change during the pulvinar inactivation. The attached figure shows what the reviewer proposed. No significant differences were found for the gamma-band oscillation in area 21a between the control and injected conditions (p>0.05). Oscillatory responses tend to decrease throughout the conditions. This reduction of activity can mean a loss of activity over time. Similar effects were observed for the other oscillations and cortical regions.

 

Figure R1
. Gamma-band frequency for area 21a considering MUAe for recovery less than for control condition. Vertical bar, interquartile range. Horizontal solid line, average. Horizontal dash line, median.

The two selection criteria classified correctly cortical VERs that change by the inactivation of the thalamus. As the new figure 5 shows, control and recovery conditions had no significant differences across cortical layers in areas 17 and 21a (compare the magnitude between the blue and black lines). Only 3 over 20 (15%) cases had significant differences between control and recovery conditions: layers IV and V for alpha waves, and layer V for gamma waves. In these cases, the recovery is considered as a partial recovery (recovery is lower than the inactivation period).

Besides, it is almost impossible to avoid the variation of the oscillatory activity, otherwise, the injection state could not be compared to any other state.

22. Fig 4 These selected responses are analyzed in Figure 4. Again they are pooling layers (which might mask effects); I see no effect, at best a tiny effect. (Again, they need to use a consistent color scheme).

We agree with the reviewer that the effect of pulvinar inactivation seems low. To quantify the variation produced by the pulvinar inactivation, we performed a new analysis that allowed us to calculate a percentage of variation. This new index was calculated by normalizing the two distributions (during and after pulvinar inactivation), by the distribution obtained before the inactivation. Such normalization allows us to quantify the impact of thalamus inactivation on corticalSuch normalization allows us to quantify the impact of thalamus inactivation on alpha and gamma-waves alpha and gamma-waves. We have added these results as an inset in Figure 4. At the population level, the effect of the inactivation of the pulvinar was low. The oscillatory cortical activity was modified by ~10%. We wanted to know if the inactivation of the thalamus affected in a homogeneous way the oscillatory activity of the cortical layers. This allowed us to perform the analysis showed in Figure 5. We have clarified this in the text.

23. Similarly in Figure 5 they show small magnitude effects, again presumably from this sample chosen with nonstationary responses as above.

As we discussed in the previous question, we added a new analysis in Figure 5. This analysis quantifies how much variation exists between the periods before and during the inactivation of the thalamus. For instance, during the LPl inactivation, alpha responses of layers II/III, IV and VI had an increase of ~84%, ~14%, and ~37% of %Var(Ctr), respectively. In area 21a, layer IV showed most of the frequency-band changes during the thalamic inactivation with an increase of ~103%.

A caveat of analyzing variations in decibels (dB) is that it uses the logarithmic scale as a reference for change. This scale is not linear. So, small negative changes in dB are highly represented, and substantial positive changes remain almost constant concerning the ratio between the experimental period and the baseline. For example, -6 dB is an equivalent to power ratio of 0.25, and 6 dB is almost ~ 4 of the power ratio. This new index of variation gives us a better quantification of the change produced by the inactivation of the pulvinar.

24. What does it mean that so many responses (power changes) were not different from baseline even in control condition? This suggests weak neuronal responses; could this for example be due to the extended grating stimuli used, which might not have been effective at driving cells with suppressive surrounds known to be prominent in visual cortex?

The reviewer is correct that the activity, in some cases, was at the baseline level. In the case of the alpha oscillations, the reviewer's suggestion is likely valid. The stimulus can generate a suppressive effect that eliminates the oscillatory response. We have argued previously that the stimulation is not the most appropriate for detecting alpha activity. We apologize for not being as precise in the description. Besides, baseline activity has some oscillatory activity that, when used as a normalization, helps to decrease the oscillatory response during the presentation of the stimulus. It should be pointed out that the animals are under anesthesia, and the oscillatory activity decreases considerably. We have addressed this issue in the Discussion.

25. Fig6 Granger Causality: Fig5 shows that in area21a alpha and gamma power did not change much in layer 5 (which provides much of the feedback signal), so why might one expect changes here in Fig6? And the authors should justify focusing on layer 5 only since cortical feedback can arise from layer 6 as well.

We would like to thank the reviewer for pointing out that the feedback connection from layer 6 of area 21a to area 17 was not analyzed. We did not concentrate our analyses on this feedback connection because it did not give positive results. We have included the analysis as a supplementary figure. The coupling index was calculated for the functional feedback connection between layer 6 of area 21a and layer 3 of area 17. For the period of 0.5 to 1 second, there is a slight difference at low frequency (0-3 Hz) between the Granger causality before and during (confidence intervals are slightly separated). However, the distributions are similar for the alpha-band frequency.

26. Were responses analyzed here the same selected (possibly nonstationary) ones shown in Figs 4 & 5?

Yes, they are the same data selected. We avoid analyzing non-stationary processes by considering 500 ms time windows (Cohen, 2014).

27. Data do not seem to match authors’ description in the text. (Starting with line 336) I do not see the 30hz peak in every condition, so the authors should be clearer.

Sorry. We have reorganized this section for clarity.

28. The authors should remind the reader what increased values of GC index mean. This is not adequately explained.

We thank the reviewer for pointing out such a lack of explanation. We have modified the first paragraph of section 3.3 to clarify and better describe the use of Granger Causality.

29. Line 349 not influences but indices. I do not understand what the authors are trying to say here.

Agreed. We have replaced the term influence by index or coupling force throughout the text.

30. Why offset of shadowed area from trace in lower left panel?

We thank the reviewer for noting this. It was a problem with the image when it was converted to EPS format. The confidence interval was aligned to the line showing the average activity.

31. The red and blue traces were not identified as control and injection conditions in either the text or figure legend.

We have included the explanation in the legend and the caption of the figure.

 

32. Discussion is overlong, and should be condensed considerably.

The discussion was reduced by two-third.

References

Berson DM, Graybiel AM. 1983. Organization of the striate-recipient zone of the cat’s lateralis posterior-pulvinar complex and its relations with the geniculostriate system. Neuroscience. 9:337–372.

Casanova C. 1993. Response properties of neurons in area-17 projecting to the striate-recipient zone of the cats lateralis posterior-pulvinar complex - comparison with cortico-tectal cells. Exp Brain Res. 96:247–259.

Cohen, M.X. The Fourier Transfor and the Convolution Theorem. In Analyzing Neural Time Series Data.

1st ed.; Cambridge, MA: MIT Press, USA, 2014; pp. 121-139.

de Souza, B.O.F.; Cortes, N.; Casanova, C. Pulvinar Modulates Contrast Responses in the Visual Cortex as

a Function of Cortical Hierarchy. Cereb Cortex. 2019, 0, 1 - 19

Hutchins B, Updyke BV. 1989. Retinotopic organization within the lateral posterior complex of the cat. J Comp Neurol. 285:350–398.

Pooresmaeili, A.; Poort, J.; Roelfsema, P.R. Simultaneous selection by object-based attention in visual and frontal cortex. Proc Natl Acad Sci U S A. 2014. 111, 6467-6472. doi: 10.1073/pnas.1316181111.

 

Rosier, A.M.; Arckens, L.; Orban, G.A.; Vandesande, F. Laminar distribution of NMDA

receptors in cat and monkey visual cortex visualized by [3H]-MK-801 binding. J Comp Neurol.

1993, 335, 369 – 380.

 

Self, MW.; Kooijmans, R.; Supèr, H.; Lamme, V.; Roelfsema, P.R. Different glutamate receptors convey feedforward and recurrent processing in macaque v1. Proc Natl Acad Sci U S A. 2012. 109, 11031–11036.

 

Self, MW.; van Kerkoerle, T.; Supèr, H.; Roelfsema, P.R. Distinct roles of the cortical layers of area V1 in figure-ground segregation. Curr Biol. 2013. 23, 2121-2129. doi: 10.1016/j.cub.2013.09.013.

 

Spaak, E.; Bonnefond, M.; Maier, A.; Leopold, D.A, Jensen O. Layer-specific entrainment of

γ-band neural activity the α rhythm in monkey visual cortex. Curr Biol. 2012, 22, 2313 – 2318.

 

Van Drongelen, W. Wavelet Analysis: Time Domain Properties. In Signal Processing for Neuroscientists.

1st ed.; Academic Press, USA, 2007; pp. 245-264.

 

van Kerkoerle, T.; Self, M.W.; Dagnino, B.; Gariel-Mathis, M.A.; Poort, J.; van der Togt, C.; Roelfsema, P.R. Alpha and gamma oscillations characterize feedback and feedforward processing in monkey visual cortex. Proc Natl Acad Sci U S A. 2014, 111, 14332 - 14441.

 

Vanni, M.P.; Casanova, C. Surround suppression maps in the cat primary visual cortex. Front Neural Circuits. 2013; 7, 1-13.

 

Xing, D.; Yeh, C.I.; Shapley, R.M. Spatial spread of the local field potential and its laminar variation in visual cortex. J Neurosci. 2009, 29, 11540–11549.

Reviewer 2 Report

By recording from primary visual cortex and area 21a in the cat, the authors demonstrate here that silencing the pulvinar leads to an enhancement of cortical oscillatory responses to visual stimuli, and that the pulvinar plays an important gating role in information routing between V1 and area 21a. These findings are important, and add to recent demonstrations of the importance of higher-order thalamus in the regulation of information flow in the cortex. However, better organization of the data and the text, and clarifications of experimental methods is required to significantly improve this manuscript-

How many animals were used for this study? Make it clear in the Methods and Results sections that V1 and area 21a were recorded simultaneously. Explain better why example data from seemingly arbitrarily chosen layers are shown in Figures 1 and 2. Explain how the LPl subdivision of the pulvinar was selectively inactivated. Do the stereotaxic coordinates used for the pulvinar injection selectively target the LPl subdivision? Please keep the colors of traces and graphs (black, blue, red) consistent throughout the manuscript. The colors in Figure 1 are inconsistent with the rest of the figures, and the legend for Figure 1 incorrectly describes the colors. Provide more quantification (p-value, number of animals/recordings) for Fig 1A. Is the increase in MUA upon thalamic inactivation consistent across recordings? How much does the MUA increase? Figures 1B and 1E show clear enhancements in oscillatory activity upon visual stimulus (even before thalamic inactivation). However, in Fig S1, the power of alpha and gamma oscillations is often not different from (and even below) baseline levels. For example, compare Fig 1E with Fig S1 C and D (for layer 4). How do the authors explain this discrepancy? Please clarify this in the main text. What do the downward arrows in Fig 1G and 1H indicate? Fig 1F suggests a clear increase in alpha oscillations in layer 4 of area 21a. Is this not a consistent or statistically significant response? Fig S1 shows that alpha oscillations are not enhanced in area 21a for layer 4. Consider using a more representative example. Label A3, B3, C3, and D3 in Figure 2. Line 254 states that “In area 21a, LPl inactivation yielded similar changes in the LFPs as those seen in area 17 (Figure 2 C and D)”. This is not really accurate since you are comparing data from different layers, and in any case is only true for low frequency oscillations. Please revise this sentence. In Figure 3B, r = 0.18 in the figure, but r = 0.17 in the main text. It is unclear to me why Figure S1 is relegated to the supplementary material, with redundant figures in Figure 5. I would recommend replacing Fig. 5 with Fig. S1. Seeing the data for all four cases, and for all layers, makes it easier for the reader to appreciate the findings. Label the red and blue traces in Figure 6 (blue: ctr, red: inj?). Also refer to the upper, middle, and bottom rows when describing Fig. 6B in the main text. For example, line 338 could read “Panel B (upper row) shows the feedforward influence before and during LPl inactivation.” Line 360: “intra- and inter-cortical”? I think simply stating “intracortical” would be more appropriate. The explanation in line 377 for the transient response in not clear to me. How does it explain the transience of the response? Line 409: “notorious”; did you mean “notable”? The use of the abbreviation “PS cells” for parvalbumin-expressing neurons is atypical. Why not PV? More detailed descriptions are required for the columns in Tables S1 to S4. The percentages in Table S1 (incorrectly titled as “Table 1”) are not intuitive. The “P-values” column in Tables S2 to S4 are also not clearly explained.

Author Response

We thank the reviewer for his/her careful reading of the manuscript and his/her suggestions. We have made all requested changes to the text, carried out additional statistical analyses and included supplementary figures.

 

1. How many animals were used for this study?

Two cats, we have added this information in the text.

 

2. Make it clear in the Methods and Results sections that V1 and area 21a were recorded simultaneously.

We have added this information to the text.

 

3. Explain better why example data from seemingly arbitrarily chosen layers are shown in Figures 1 and 2.

This is now explained in the Result section.

 

4. Explain how the LPl subdivision of the pulvinar was selectively inactivated. Do the stereotaxic coordinates used for the pulvinar injection selectively target the LPl subdivision?

The validation of GABA injection in the LP nucleus was performed by a series of coronal sections of the cat thalamus stained for Acetylcholinesterase showing the spread of the Chicago Sky Blue dye in the lateral subdivision of LP. Moreover, recordings were simultaneously in that region to quantify whether the inactivation was successful. After cessation of the injection, a recovery of neuronal activity was observed. Besides, we estimated the LP scotoma by comparing the extent of the Chicago Sky Blue staining with the retinotopic maps described by Hutchins and Updyke (1989) as we did in de Souza et al., 2019. A new figure has been added in the supplementary material.

 

5. Please keep the colors of traces and graphs (black, blue, red) consistent throughout the manuscript. The colors in Figure 1 are inconsistent with the rest of the figures, and the legend for Figure 1 incorrectly describes the colors.

Corrected. We used the same colours consistently throughout the article.

 

6. Provide more quantification (p-value, number of animals/recordings) for Fig 1A.

We apologize for the poor quality of Figure 1. We have improved this figure and its explanation. We have added more information about the statistical results.

 

7. Is the increase in MUA upon thalamic inactivation consistent across recordings? How much does the MUA increase?

We have calculated the percentage of variation for the periods during and after the injection. These percentages have been added to Figure 1.

 

 

8. Figures 1B and 1E show clear enhancements in oscillatory activity upon visual stimulus (even before thalamic inactivation). However, in Fig S1, the power of alpha and gamma oscillations is often not different from (and even below) baseline levels. For example, compare Fig 1E with Fig S1 C and D (for layer 4). How do the authors explain this discrepancy? Please clarify this in the main text.

Following the reviewer's suggestions, we have changed the examples shown in Figure 1. We now present more examples of LFPs, and we changed the scale of the FFT to describe more significant differences between conditions. Furthermore, we have explained in the Discussion why the values are at the baseline level.

 

 

9. What do the downward arrows in Fig 1G and 1H indicate?

They showed the co-activation of the LFP and MUA for gamma waves. It is now explained.

 

10. Fig 1F suggests a clear increase in alpha oscillations in layer 4 of area 21a. Is this not a consistent or statistically significant response? Fig S1 shows that alpha oscillations are not enhanced in area 21a for layer 4. Consider using a more representative example.

We have changed the representative example, as indicated in the answer to question 8.

 

Label A3, B3, C3, and D3 in Figure 2.

We added such indexes to make the figure more accessible to the reader.

 

Line 254 states that “In area 21a, LPl inactivation yielded similar changes in the LFPs as those seen in area 17 (Figure 2 C and D)”. This is not really accurate since you are comparing data from different layers, and in any case is only true for low frequency oscillations. Please revise this sentence.

We have rewritten a significant portion of these sections to answer this concern.

 

In Figure 3B, r = 0.18 in the figure, but r = 0.17 in the main text.

We have replaced it for r=0.18.

 

It is unclear to me why Figure S1 is relegated to the supplementary material, with redundant figures in Figure 5. I would recommend replacing Fig. 5 with Fig. S1. Seeing the data for all four cases, and for all layers, makes it easier for the reader to appreciate the findings.

We have changed Figure S1 by Fig. 5. We have added, nonetheless, a percentage of variation to quantify such variations. This new index shows that there are consistent variations in the layers that increased significantly due to the inactivation of the thalamus.

 

Label the red and blue traces in Figure 6 (blue: ctr, red: inj?). Also refer to the upper, middle, and bottom rows when describing Fig. 6B in the main text. For example, line 338 could read “Panel B (upper row) shows the feedforward influence before and during LPl inactivation.”

We are grateful for the suggestions; we have updated the figures accordingly.

 

Line 360: “intra- and inter-cortical”? I think simply stating “intracortical” would be more appropriate.

We agree and make the change throughout the text.

 

The explanation in line 377 for the transient response in not clear to me. How does it explain the transience of the response?

We are very sorry for missing this point. We have rearranged the explanation and added another paragraph to explain why this transient activity for alpha waves exists.

 

Line 409: “notorious”; did you mean “notable”?

Corrected.

 

The use of the abbreviation “PS cells” for parvalbumin-expressing neurons is atypical. Why not PV?

Corrected.

 

More detailed descriptions are required for the columns in Tables S1 to S4. The percentages in Table S1 (incorrectly titled as “Table 1”) are not intuitive. The “P-values” column in Tables S2 to S4 are also not clearly explained.

We have included more information in the supplementary material. In fact, we have improved the description of all tables presented in the supplementary material.

 

 

References

 

Hutchins B, Updyke BV. 1989. Retinotopic organization within the lateral posterior complex of the cat. J Comp Neurol. 285:350–398.

Round 2

Reviewer 1 Report

The authors have addressed the issues I raised in my previous review, and the manuscript is much improved. I raise a few minor points that the authors should address.

1) Abstract line 21 “In area 21a, gamma oscillations, except for layer I, had more significant responses…”   Authors should rephrase – how can oscillations have responses?

2) Figure 1 panels I, J should be H, I to match text.

And the legend states “In panels A, C, D and E, shadow areas represented the bootstrapping of random averages (1,000 repetitions), to obtain a confidence interval at a threshold P=0.05.”     Do they mean panels A, C, D, and *F*?     The authors need to be wary of such careless mistakes that confuse the reader. They should recheck that all text and legend descriptions match their figures.

3) Figure 5: Why don’t the blue traces (control condition) all equal ZERO, i.e. no change from baseline? Surely that is what control means? A brief reminder for the reader here would be useful to clarify interpretation of this figure.

For example in panel D1, there is a significant effect in layer 6, but the injection (red) value is near 0dB change from baseline, while the CONTROL value is roughly -2dB change from baseline. How can control differ from baseline? Or a significant effect be 0dB change from baseline?

4) Fig5 legend states “Symbols in Figure 1 only are for significant differences…”   Do the authors mean panels A1, B1, C1, D1? Not figure 1?...

5) Fig S1 legend line 4 should say custom BUILT injectrode (not build)

Author Response

1) Abstract line 21 “In area 21a, gamma oscillations, except for layer I, had more significant responses…” Authors should rephrase – how can oscillations have responses?

We have rephrased line 21, as suggested.

 

2) Figure 1 panels I, J should be H, I to match text.

And the legend states “In panels A, C, D and E, shadow areas represented the bootstrapping of random averages (1,000 repetitions), to obtain a confidence interval at a threshold P=0.05.”    Do they mean panels A, C, D, and *F*? The authors need to be wary of such careless mistakes that confuse the reader. They should recheck that all text and legend descriptions match their figures.

We have corrected and matched all letters in Figure 1.

 

3) Figure 5: Why don’t the blue traces (control condition) all equal ZERO, i.e. no change from baseline? Surely that is what control means? A brief reminder for the reader here would be useful to clarify interpretation of this figure. 

We understand where the confusion comes from. In fact, the “control” condition was defined as the cortical oscillatory activity in response to the drifting gratings and not as a “baseline” condition where there is no visual stimulation. We agree that the results shown in figure 5 would be better appreciated with a more precise definition of “control.” Thus, we modified the figure legend to include a more detailed description of the parameter “dB change from Baseline/s.”

 

For example in panel D1, there is a significant effect in layer 6, but the injection (red) value is near 0dB change from baseline, while the CONTROL value is roughly -2dB change from baseline.

What we want to emphasize here are the differences between conditions. In particular, the differences between the control (blue line) and the injection periods (red line). The dotted black lines at dB = 0 are used only to better show the direction of the change (increase vs. decrease of LFP amplitude).

 

How can control differ from baseline?

The "control" condition differs from the baseline because the visual stimulation causes changes in the oscillations. These changes may vary depending on the frequency bands and the layers analyzed.

 

Or a significant effect be 0dB change from baseline?

No. A significant effect is the difference in" dB change/s" between blue (Ctrl) and red (Inj) lines.

 

4) Fig5 legend tates “Symbols in Figure 1 only are for significant differences…”  Do the authors mean panels A1, B1, C1, D1? Not figure 1?...

We agreed. We now use panels A1, B1, C1, and D1.

 

5) Fig S1 legend line 4 should say custom BUILT injectrode (not build)

Corrected.