Worry Modifies the Relationship between Locus Coeruleus Activity and Emotional Mnemonic Discrimination

Round 1

Reviewer 1 Report

In the legend of Figure 5 'to' should be 'too'.

Author Response

We thank the reviewers for reading through the manuscript and their feedback.

Comments and Suggestions for Authors. In the legend of Figure 5 'to' should be 'too'.

Our reply: we apologise for this oversight and have corrected this mistake. on page 13 in figure 5

Reviewer 2 Report

I would like to congrats with the authors for this high quality and extremely interesting article. In my opinion the research has been conducted meticulously and is supported by a solid rationale as well as a well-selected bibliography. I also agree with the authors’ point of view about the relationship between the memory performance that needs a balanced LC activity to be optimized (or rather to not be affected by an unbalanced LC activity). However, it must be noticed that no mention has been made about the possible effect of an unbalanced activity of the LCs of the two sides. As this point has been observed to have strong influences on the cognitive performances and behavior, this could probably explain some of the observed outcomes (https://doi.org/10.3389/fnsys.2021.749444 ; https://doi.org/10.3389/fnana.2017.00130).

Finally, I suggest to the authors to evaluate the pupillometry as a reliable and chip test to evaluate the LC activity for their future studies.

A short-note regarding the Figure S1 (mentioned at line 115) that has not been provided among the submitted material.

Author Response

We thank the reviewers for their enthusiasm about our manuscript and for their remarks and the opportunity to improve our manuscript. Below we have added our response to your suggestions. 

I would like to congrats with the authors for this high quality and extremely interesting article. In my opinion the research has been conducted meticulously and is supported by a solid rationale as well as a well-selected bibliography. I also agree with the authors’ point of view about the relationship between the memory performance that needs a balanced LC activity to be optimized (or rather to not be affected by an unbalanced LC activity). However, it must be noticed that no mention has been made about the possible effect of an unbalanced activity of the LCs of the two sides. As this point has been observed to have strong influences on the cognitive performances and behavior, this could probably explain some of the observed outcomes (https://doi.org/10.3389/fnsys.2021.749444 ; https://doi.org/10.3389/fnana.2017.00130).

Our reply: Thank you for your comment. In the current study we have averaged the left and right LC as well as the left and right hemispheres, as we had no a-priori hypothesis on possible lateralization. However, it would be interesting for future studies to further investigate the contribution of the two sides. We have therefore added the following to our limitation section on page 15 line 437-440:
Although we averaged over the left and right hemisphere future studies should consider further exploring a disbalance in activity between the two LC’s and hemisphere as this has been shown to be related to cognitive functioning and could thereby further support our results [70, 71].

Finally, I suggest to the authors to evaluate the pupillometry as a reliable and chip test to evaluate the LC activity for their future studies.

Our reply: We thank the reviewer for their suggestion. We agree that pupillometry would allow us in future studies to further understand the relation between a disbalance in LC activity and cognitive performance. we therefore have suggested to add pupil measurements to future studies to provide information about the tonic and phasic activity of the LC during tasks as BOLD alone would not provide us this information. see page 14 line 399-405.  
However, to really understand the Yerkes-Dodson relationship between LC activity and memory performance, and the potential modification by worry, additional measurements sensitive to the disbalance between tonic and phasic activity of the LC, such as pupil measurements, are needed as BOLD cannot reveal us information about phasic activity [5]. This could provide us further information on how the Yerkes-Dodson relationship between LC activity and memory performance changes under high worry and high arousal in older adults.

A short-note regarding the Figure S1 (mentioned at line 115) that has not been provided among the submitted material.

Our reply: We apologise for this oversight and have checked the supplementary materials for figure S1 to make sure the figure is included on page 2 of the supplementary materials

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

General Comments:

The study examines the role of worry and arousal in the relationship between the locus coeruleus and emotional memory function. 

In general the manuscript would benefit from a general proof-reading for grammatical and spelling errors/word choice. I have noted several of them but as a manuscript should generally be proofread by the author before submission, and it is not a reviewer’s job in general to provide a proofreading service, I have left the rest.

The clarity of the ideas presented is somewhat lacking. Important ideas are not explained fully, or examined in a depth that provides a proper context for understanding the motivation, methods, analysis, and conclusions of the paper. Further, and importantly, some crucial references are not provided.

I see no mention in the Discussion section as to why 

Specific Comments:

l.38: all emotions or only specific ones? Emotion is an extremely general term. Expand this and provide references.

L.39: not solely from the LC. see noradrenergic cell groups (A1-A7). Amend this statement as it is misleading to the reader.

Ll.43-44: “Interestingly, both the LC and the medial temporal lobe (MTL) play an important role 43 in both memory performance and emotional regulation…” - please provide references for this claim.

Ll.44-45: …”two processes that are well known to be affected by ageing…” - again, reference this.

L.50: “hippocampal pattern separation” - what is this?? And how is it relevant to this study?

L.51: MTL, not MLT

l. 51: "Within the MLT this process relies on the Dentate Gyrus…” - which process?? Hippocampal pattern separation or mnemonic discrimination? 

l.63: remove reference text

L.67: “…individuals often report worries about their memory functioning, and as such may at…” - may BE at

l.100-101:  what exactly was the sampling procedure? I think this should be included in the main text. As I do not have access to the supplementary materials I am left wondering if it was random within these groups or some other arbitrary method was used?

L.130-135: So you are assuming these pictures affect arousal in a particular way. This should be noted. What is the reasoning/research behind this assumption? This should also be noted. You include no independent measure of arousal. Was physiological data collected? Was it possible to derive an objective measure of arousal? If so, why was this not done? This is an extremely important question as both your hypothesis and conclusions are dependent on it.

L.135: “objects” not “object”

L.135: a period is required between sentences.

l.137: please not if anything was done to ensure attention was on the preceding arousal-inducing stimulus as well. If not, why not?

l.142: please state the length of this resting period in the main text. Also, what were the instructions to the participant during this period? Were their eyes opened or closed? And was there a delay between scans, and if so how long?

Ll.150-152: please clarify what "preceding high and low arousal context" means. I'm trying to think whether it could be the previous item in the recall stream of stimuli or the previous item to the target item in the encoding stream or something else entirely. It is impossible to interpret the following findings without knowing this. Both would presumably have influences on arousal?

Ll.211-220: was any physiological correction (e.g., RETROICOR) performed? Arousal and anxiety are generally accompanied by increased cardiac and respiratory rates, changes in HRV, and changes in respiration volume per time (RVT) are also known to cause large changes in signal intensity, for example. If this was not or will not be done, this needs to be at the very least acknowledged and justified either here in the Methods section or later on in Study Limitations.

ll.222-224: there is an emphasis in the introduction (it’s even mentioned in the abstract). Why was an ROI analysis not performed on the MTL?

ll.231-233: please see comment for ll.211-220

l.243: “robust multiple linear regression: Please justify this choice further, as it is somewhat unusual, and explain in detail the particular method of robust regression used, the estimator, scaling function, etc. Was the scaling factor calculated iteratively or was it predetermined? Some of the findings in Figure 2 appear to be driven by outliers, regardless of having used a robust method. More details are required.

l.294: “relationship” rather than “relation”

L.299: “worry”, not “worries”

L.300: “contrast” not “contract”

l.302-304: this sentence is not clear. Possibly a grammatical issue?

l.316 (figure 4): In general I find this figure extremely confusing. As a reader I am left guessing what any of this means. It is not clear, either in the main text or the figure sub-text what exactly was done for this analysis or what any of the dots or lines represent. This most definitely needs to be re-worked: the main text (p.9), the caption sub-text, as well as the figure itself. Also, are these values normalized? I'm assuming the y-axis is. What about the x-axis? Also, why is the lower legend continuous if the variable is represented categorically? The colors of the dots are also obscured by the confidence interval overlays - there seems to be more than 3 colors of dots also. What do the p values actually represent? Please reverse the top right legend so it is consistent with the lower one….

l.322: “complaint” not “complain”

l.326: “…had a beneficial effect…” - it was "correlated with", there is no causal or directional inference possible here

l.328-329: there is no younger control group in this study, so to claim that this is the case for "older individuals" is possibly misleading. It is the case for all ages of people and for animals that noradrenaline need to be regulated properly for optimal attentional function. There is a fair body of research addressing this question.

Ll.361-366: It would be interesting here to the reader I think to include some speculation on possible reasons why LC activity had no effect on memory in the high-worry group. Are they known to have higher or lower general arousal? Could there be some ceiling effect with LC activity? Or are there worry and anxiety engaging subsystems of attention and thus not allowing them to modulate LC effectively? Some circumspection here would be good.

l.401: I would revisit this section on Study Limitations in the context of the above comments.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

In their submission “Worry modifies the relationship between locus coeruleus activity and emotional mnemonic discrimination” Pagan et al. examine the relationship between memory worry and similar pattern discrimination memory performance while capturing brain activity in memory-implicated brains regions, the LC, hippocampal subfields, and the amygdala, assessed using a 7T fMRI scan during encoding. Using a previously established emotional discrimination paradigm they assessed memory performance under both a lower and a higher arousal condition in a group of well-characterized subjects from the Maastricht aging study.  They selected two subsets of individuals who were consistently classified as having low or high worries about mnemonic abilities for participation. The data were analyzed in linear regression models to capture the best fit for the modulatory effect of worry under low and high arousal conditions indexed by behavioral task context.  Given the care in experimental design, their significant results are of particular interest. They replicate an earlier finding that higher activity in the dentate gyrus is associated with more accurate discrimination of similar items. While they do not support their initial prediction that higher LC activity will characteristically be associated with more accurate discrimination, the data are consistent with an inverted U curve relationship between LC activity and improved mnemonic performance. With aging a requirement for optimal LC activity levels may be even more acute than in younger individuals.  In the present study higher LC activity is associated with more accurate discrimination memory in low worriers while higher LC activity does not improve performance in high worriers. These are very interesting outcomes and are carefully discussed. The need for future studies that may offer the possibility of probing the patterns of LC activity during encoding is highlighted.  The translational implications of this work are of particular interest.

Minor editing suggestions for the authors.

1. 3 neutral daily objects (rather than object)
2. 14 complaint (rather than complain)
3. 15 was there (rather than there was)

p.15 we visualized the LC at an individual (rather than individual)

Author Response

Brainscience 1546414

Reviewer 2

Comments and Suggestions for Authors

In their submission “Worry modifies the relationship between locus coeruleus activity and emotional mnemonic discrimination” Pagan et al. examine the relationship between memory worry and similar pattern discrimination memory performance while capturing brain activity in memory-implicated brains regions, the LC, hippocampal subfields, and the amygdala, assessed using a 7T fMRI scan during encoding. Using a previously established emotional discrimination paradigm they assessed memory performance under both a lower and a higher arousal condition in a group of well-characterized subjects from the Maastricht aging study.  They selected two subsets of individuals who were consistently classified as having low or high worries about mnemonic abilities for participation. The data were analyzed in linear regression models to capture the best fit for the modulatory effect of worry under low and high arousal conditions indexed by behavioral task context.  Given the care in experimental design, their significant results are of particular interest. They replicate an earlier finding that higher activity in the dentate gyrus is associated with more accurate discrimination of similar items. While they do not support their initial prediction that higher LC activity will characteristically be associated with more accurate discrimination, the data are consistent with an inverted U curve relationship between LC activity and improved mnemonic performance. With aging a requirement for optimal LC activity levels may be even more acute than in younger individuals.  In the present study higher LC activity is associated with more accurate discrimination memory in low worriers while higher LC activity does not improve performance in high worriers. These are very interesting outcomes and are carefully discussed. The need for future studies that may offer the possibility of probing the patterns of LC activity during encoding is highlighted.  The translational implications of this work are of particular interest.

 We thank the reviewer for their enthusiasm about our manuscript and their thorough and excellent remarks and the opportunity to improve our manuscript. Changes to address each of your concerns are highlighted in red in the manuscript. 

Minor editing suggestions for the authors.

1. 3 neutral daily objects (rather than object)
2. 14 complaint (rather than complain)
3. 15 was there (rather than there was)

Our reply: Thank you for pointing out these spelling mistakes to us. We have changed all your suggestion in our manuscript.

p.15 we visualized the LC at an individual (rather than individual)

Our reply: We apologize for this oversight and have changed the sentence in line with your comment.

We have added the adjusted manuscript in the attachment. 

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript investigates the noradrenergic system activity with respect to emotional mnemonic discrimination in healthy older adults with a high or low worry index. The study is fundamentally well structured, and manuscript well written. There are, however, some issues that should be addressed:

1. I am slightly concerned about the measurement of worry in this study. The authors state they assessed ‘worry’ from a known and used questionnaire (the Metamemory in adulthood (MIA) questionnaire). This questionnaire does not have a worry-score per se but the authors used the MIA-anxiety score for the purpose. From G Davey (1992) who compares several worry questionnaires, one of the main reports is that worry and anxiety are separate constructs, and that there are different features of worry that are captured by different questionnaires. The authors should recognize that the study might not be assessing worry effects on its own just using the MIA-anxiety score. Would also be useful to know the absolutely range of the MIA-anxiety score.
2. How much B0 dropout was there around the middle-temporal lobe with the forward tilt of the fMRI slab? Did it affect the hippocampus fMRI time-series in any of the subjects? How much distortion was observed in this region? Could the authors provide an example of the distorted and distortion corrected fMRI volume?
3. Paragraph starting in line 181: were manual corrections performed on the entire white matter mask, and were the autorecon2 and autorecon3 processes from FreeSurfer ran again after the manual corrections?
4. LC segmentation: It would be helpful for the reader to include a figure of the final segmentation of what the authors considered to be the LC region-of-interest (ROI) in the template space. Considering that it is a manual delineation it is always biased towards a specific rater. When the mask is then mapped to individual space, a transform is applied. The binary mask suffered interpolation so it would be helpful to state what type of criteria did the authors use to define the individual’s LC ROI.
5. Paragraph starting on line 222: To find the transformation between the T2* based EPI to T1 space, it has been shown that the boundary based registration is more robust than cross-correlation or normalized mutual information methods and is considerably more robust when there are strong intensity inhomogeneities (Greve and Fischl. Neuroimage. 2009). In addition, it was important for correction of small deviations at submillimetre level (Huang et al. Neuroimage. 2020). Considering registrations like for the LC mask that are essential be as accurate as possible, I was just wondering if the authors evaluated the registrations, and if the BBR method would be most appropriate. It would also be helpful to know the details of this 12-DOF registration.
6. Lines 231-235: How did the authors correct for 4th ventricle signal? Was it manually segmented or was it defined by an Atlas? It isn’t clear if the authors just regressed the temporal derivates of the white-matter signal, 4th ventricle signal, motion parameters and motion outliers, or if they also included the regression of e.g. the global mean or first principle components of these variables.
7. Did the authors record any external physiological features like recordings from the pulse oximeter and the respiratory bellows? If these are available, why weren’t they included as covariates of no interest to denoise the data?
8. To my understanding, and although the authors point to a number of figures in supplementary material, this was not provided for reviewing.
9. Figure 4: Although representing an interaction effect is sometimes complicated, I find this plot very confusing specially when the authors try to stratify worry scores into three categories by plotting three regression lines. If the authors assume this stratification (-1SD, mean, +1SD) how many subjects are included in each category? Would it be possible to fit simpler sub-models in each category? Maybe this would enhance the effects that are observed, as for one of the three plots (Figure 4A) is not significant after multiple comparison correction, and another is almost falling above the p=0.05 threshold of significance (Figure 4C).
10. Would it be helpful to include a summary schematic of the interactions and interpretations that the authors make? For example paragraph starting on line 335: in memory performance, the LC and amygdala not seen directly linked during high arousal. But the authors hypothesise that it might be mediated by the CA1.

Some minor issues:

1. Abstract should be carefully formatted.
2. Lines 251,252: Should be ‘Memory Performance ~’ instead of ‘Memory ~’
3. Figure 2 caption uses a number of abbreviations but the main figure has the words in full. Would be good to keep consistency and define all abbreviations including IQR.
4. There must be a typo in the voxel size of the fMRI scans as with a 0.125mm slice thickness, 50 slices the acquisition would only cover 6.25mm in the slice direction (line 172).

References:

1. Davey, G.C., 1993. A comparison of three worry questionnaires. Behaviour research and therapy, 31(1), pp.51-56.
2. Huang, P., Carlin, J.D., Henson, R.N. and Correia, M.M., 2020. Improved motion correction of submillimetre 7T fMRI time series with Boundary-Based Registration (BBR). NeuroImage, 210, p.116542.
3. Greve, D.N. and Fischl, B., 2009. Accurate and robust brain image alignment using boundary-based registration. Neuroimage, 48(1), pp.63-72.

Author Response

Please see the attachment

Author Response File: Author Response.docx