Average functional gradients of the locus coeruleus in the Cam-CAN 3T and HCP 7T datasets.

A) Average functional gradients from subjects below 40 years of age in the Cam-CAN and HCP 7T datasets were projected to a surface rendering of the LC ROI mask. We present younger participants from the Cam-CAN cohort to make it comparable to the demographic characteristics of the HCP cohort. The color bar denotes normalized values of the dominant eigenmap. B) Projection maps of the LC were calculated as the mode of maximal correlation localization to either the rostral (red) or the caudal (blue) part of the LC in order to show the two extremes of connectivity. Projection maps were overlaid on the average HCP cortical surface and MNI152 template for the subcortical structures.

Demographic features of the Cam-CAN and HCP 7T cohorts.

Values in the table represent means followed by (standard deviations), except when stated otherwise. Measures from the emotional memory task represent d prime scores, which demonstrate the sensitivity of an individual to detect or correctly categorize stimuli in the given condition. Emotion regulation measures represent rescaled Likert-scale ratings, whereas the Hotel-task is scored as the average time in seconds a participant spends on each subtask. PSQI is a self-reported measure of sleep quality. HADS is a screening tool for anxiety and depression symptoms. Abbreviations: SD = standard deviation, M=male, F=female, PSQI=Pittsburgh Sleep Quality Index, HADS=Hospital Anxiety and Depression Scale.

Age-related changes in the locus coeruleus functional gradient.

Average functional gradients from representative age groups (20-30, 30-50, 50<) were projected to a surface rendering of the left and right LC ROI mask (A). The scatter plots show the relationship between age and rank-transformed values of TSM parameters (after adjusting for effects of sex) describing the spatial features of the gradient (see Methods for more details) with a least-squares line in red (B). Parameter changes correspond to a loss of rostral-like connectivity, increased asymmetry and more clustered functional organization of the LC in older participants.

Association of LC functional gradient spatial features and behavioral measures.

The polar plot depicts additional variance explained by spatial features of the LC functional gradient over and above nuisance variables (age, sex, education for the analyses to predict cognitive variables) in emotional memory, emotion regulation, executive function and sleep quality in terms of adjusted partial R2 values. Example window-averaged gradients from participants with lowest and highest emotional memory scores and emotion regulation scores were projected to a surface rendering of the LC ROI mask, showing that old-like functional gradient spatial features (less extensive rostral-like connectivity and more clustered functional organization) are associated with lower scores in emotional memory and emotion regulation related to stimuli with negative valence. No significant associations were found with executive function and sleep quality. Abbreviations: EM=emotional memory, ER=emotion regulation, PSQI=Pittsburgh Sleep Quality Index.

Alterations of the LC functional gradient in participants with abnormal HADS scores.

The violin plots depict gradient spatial model parameter distributions in the two groups. Example gradients from participants with low and high HADS scores were projected to a surface rendering of the LC ROI mask, showing that high HADS scores were associated with a reduction of rostral-like connectivity, as well as a greater asymmetry in the spatial layout of the bilateral LC gradients.

Correlation plots of individual trend surface model parameters and cognitive scores.

The scatter plots depict residuals of rank-transformed gradient parameters and cognitive scores after correcting for age, sex and education, with a least squares line superimposed in red color.

Normalized difference map depicting the similarity of the gradient between subjects below 40 years of age from the CamCAN and the HCP 7T dataset.

We present younger participants from the Cam-CAN cohort to make it comparable to the demographic characteristics of the HCP cohort. The difference map was calculated as the absolute difference by subtracting the normalized CamCAN gradient from the normalized HCP gradient.

Average functional gradients of the locus coeruleus in the Cam-CAN 3T and HCP 7T datasets using a LC ROI different from the main analysis. Average functional gradients from subjects below 40 years of age in the Cam-CAN and HCP 7T datasets were projected to a surface rendering of the LC ROI mask. We present younger participants from the Cam-CAN cohort to make it comparable to the demographic characteristics of the HCP cohort. The color bar denotes normalized values of the dominant eigenmap.