Heritability and cross-species comparisons of human cortical functional organization asymmetry

  1. Bin Wan  Is a corresponding author
  2. Şeyma Bayrak
  3. Ting Xu
  4. H Lina Schaare
  5. Richard AI Bethlehem
  6. Boris C Bernhardt
  7. Sofie L Valk  Is a corresponding author
  1. Otto Hahn Group Cognitive Neurogenetics, Max Planck Institute for Human Cognitive and Brain Sciences, Germany
  2. International Max Planck Research School on Neuroscience of Communication: Function, Structure, and Plasticity (IMPRS NeuroCom), Germany
  3. Department of Cognitive Neurology, University Hospital Leipzig and Faculty of Medicine, University of Leipzig, Germany
  4. Institute of Neuroscience and Medicine (INM-7: Brain and Behavior), Research Centre Jülich, Germany
  5. Center for the Developing Brain, Child Mind Institute, United States
  6. Department of Psychiatry, University of Cambridge, United Kingdom
  7. McConnell Brain Imaging Centre, Montréal Neurological Institute and Hospital, McGill University, Canada
  8. Institute of Systems Neuroscience, Heinrich Heine University Düsseldorf, Germany

Abstract

The human cerebral cortex is symmetrically organized along large-scale axes but also presents inter-hemispheric differences in structure and function. The quantified contralateral homologous difference, that is asymmetry, is a key feature of the human brain left-right axis supporting functional processes, such as language. Here, we assessed whether the asymmetry of cortical functional organization is heritable and phylogenetically conserved between humans and macaques. Our findings indicate asymmetric organization along an axis describing a functional trajectory from perceptual/action to abstract cognition. Whereas language network showed leftward asymmetric organization, frontoparietal network showed rightward asymmetric organization in humans. These asymmetries were heritable in humans and showed a similar spatial distribution with macaques, in the case of intra-hemispheric asymmetry of functional hierarchy. This suggests (phylo)genetic conservation. However, both language and frontoparietal networks showed a qualitatively larger asymmetry in humans relative to macaques. Overall, our findings suggest a genetic basis for asymmetry in intrinsic functional organization, linked to higher order cognitive functions uniquely developed in humans.

Editor's evaluation

This is a valuable paper investigating hemispheric asymmetries in brain functional connectivity. The authors quantify this asymmetry using a solid methodology that capitalises on recent developments in functional gradients, and they further ask if these asymmetries are heritable and how they compare between humans and macaque monkeys. The results suggest a genetic underpinning of brain functional asymmetry, particularly in areas supporting unique human functions. These findings may help further our understanding of brain asymmetries.

https://doi.org/10.7554/eLife.77215.sa0

Introduction

The human cerebral cortex consists of two hemispheres that are not exactly alike and show marked differences in structure and function along a left-to-right axis (Geschwind and Levitsky, 1968; Güntürkün et al., 2020; Karolis et al., 2019; Kong et al., 2018; Kong et al., 2022; Liang et al., 2021; Raemaekers et al., 2018; Sha et al., 2021; Zhong et al., 2021). It has been suggested that the brain favors asymmetry to avoid duplication of neural circuitry having equivalent functions (Karolis et al., 2019; Levy, 1977). For example, bilateral cortical regions showing asymmetry in task-evoked activity have reduced (long-range) connections with the opposite homologous regions, favoring more local connectivity (Karolis et al., 2019).

Asymmetry, that is quantitative hemispheric differences between contralateral homologous regions, supports partly differentiable functional processes (Karolis et al., 2019; Galaburda et al., 1990; Call et al., 2017). Previous work has suggested that functions related to leftward dominance include language processing (Lane et al., 2017; Piervincenzi et al., 2016; Wang et al., 2019), letter search (Pollack and Price, 2019), and analogical reasoning (Urbanski et al., 2016). On the other hand, rightward dominance of functional activation has been related to holistic word processing (Ventura et al., 2019), visuospatial abilities (Chen et al., 2019), emotional processing (Moeck et al., 2020), as well as with psychiatric disorders such as autism spectrum disorder (Floris et al., 2016). In addition to task-related asymmetries, resting state functional connectivity (FC) studies have also reported hemispheric differences. For example, language areas of the middle and superior temporal cortex showed increased connectivity with regions in the left hemisphere relative to their right hemispheric counterparts (Raemaekers et al., 2018), and the right amygdala showed higher connectivity with the entire cortex than the left amygdala (Tetereva et al., 2020). Moreover, previous work has indicated that there are inter- and intra-hemispheric differences in functional connectivity between healthy adults and patients with schizophrenia (Agcaoglu et al., 2018), and between neurotypical individuals and those diagnosed with autism spectrum disorder (Hahamy et al., 2015). It is possible that such functional processing asymmetries may be driven by subtle differences in functional organization between the hemispheres.

One appealing approach to studying functional organization is by evaluating the low-dimensional axes, or gradients, present within the connectome. These approaches embed brain regions on a continuous data-driven space based on their functional connectome (Coifman et al., 2005; Margulies et al., 2016; Vos de Wael et al., 2020). Gradients capture how connectivity profiles from distinct cortical regions are integrated (i.e. similar functional connectivity profiles) and segregated (i.e. dissimilar functional connectivity profiles) across the cortex (Margulies et al., 2016; Bethlehem et al., 2020; Haak and Beckmann, 2020; Paquola et al., 2019). Regions that have similar connectivity profiles are at similar positions along these gradients, whereas regions with dissimilar connectivity profiles are placed further apart. The principal functional gradient, partly reflected in the intrinsic geometry of the cortex, shows that regions of the transmodal systems occupy locations equidistant from unimodal systems (Margulies et al., 2016; Hong et al., 2020; Murphy et al., 2018). Gradients provide a synoptic framework to capture smooth variations of connectivity patterns across the cortical mantle. They describe variations in genetic patterning (Vainik et al., 2020; Valk et al., 2020; Valk et al., 2022, functional processes Margulies et al., 2016; Murphy et al., 2018; Turnbull et al., 2020), and are observed across species (Valk et al., 2020; Coletta et al., 2020; Xu et al., 2020). Gradients have been linked to graph-theoretical markers such as degree centrality (Hong et al., 2019) and microcircuit dynamics (Park et al., 2021) as well as connectivity distance (Hong et al., 2019; Wang et al., 2021). Moreover, the principal gradient describes the geodesic distance between primary and default regions, and relates to cortical microstructure and associated processing hierarchies (Huntenburg et al., 2018). In doing so, and in contrast to clustering or network-based approaches, the gradient framework provides a spatial ordering of functional brain networks, placing them along a gradual axis of connectivity variation reaching from sensory to transmodal areas. In the context of asymmetry of gradient loadings this would mean that a given region with a significant left-ward asymmetry along the first gradient (sensory-to-transmodal) has a connectivity profile more similar to the transmodal anchor in the left hemisphere relative to the right. Consequently, these regions are placed at different positions along the cortical hierarchy, providing novel insights concerning the system-level variations in the asymmetric brain. Indeed, recent research suggests that the principal gradient is asymmetric (Liang et al., 2021; Gonzalez Alam et al., 2022) and that the degree of asymmetry relates to individual differences in semantic performance and visual reasoning (Gonzalez Alam et al., 2022).

As inter-hemispheric asymmetry has been observed consistently in human brain structure and function, there may be important (phylo)genetic factors supporting lateralized human cognition (Güntürkün et al., 2020; Corballis, 1989; Corballis, 2009; Güntürkün and Ocklenburg, 2017; Vilain et al., 2011; Vallortigara et al., 1999). Previous work has reported that brain structure asymmetry is heritable (Sha et al., 2021; Zhong et al., 2021), especially in the language areas, and differentiates between humans and non-human primates (Eichert et al., 2020; Eichert et al., 2019; Neubauer et al., 2020; Spocter et al., 2020). At the same time, it has been shown that both humans and apes show asymmetry of brain shape (Neubauer et al., 2020), indicating that asymmetry is not a uniquely human brain feature. However, asymmetry was observed to be more local and variable in humans, potentially suggesting that individual variation in asymmetry in humans varies as a function of localized networks rather than global features. It is of note that the full FC matrix contains both intra-hemispheric and inter-hemispheric connections. Intra-hemispheric connections, compared to the inter-hemispheric connections, have been suggested to reflect inhibition of the corpus callosum and may underlie hemispheric specializations involving language, reasoning, and attention (Gazzaniga, 2000). Conversely, inter-hemispheric connectivity may reflect information transfer between hemispheres, for example, of motoric information, or crude information concerning spatial locations (Gazzaniga, 2000). Previous studies have mainly employed intra-hemispheric FC to study gradient asymmetry (Liang et al., 2021; Gonzalez Alam et al., 2022). However, inter-hemispheric differences in functional connectivity may also have functional relevance. For example, inter-hemispheric connectivity has been reported to be abnormal in patients with schizophrenia (Agcaoglu et al., 2018; Chang et al., 2019) and autism (Hahamy et al., 2015). Indeed differences not only of functional organization within each hemisphere but also between hemispheres, enabled by the corpus callosum, are relevant for integration and segregation of cognitive function and support hemispheric coordination (Gazzaniga, 2000; Toga and Thompson, 2003).

Here, we investigated the genetic basis of asymmetry of functional organization. We first examined whether inter-individual differences in asymmetry of functional organization are under genetic control, that is heritable. Second, we investigated whether asymmetry of functional organization is phylogenetically conserved in macaques. To probe individual variation in asymmetry of functional organization, we utilized a data-driven nonlinear dimension reduction technique, as this approach can provide reliable and robust indices of individual variation of cortical organization (Hong et al., 2020). We first obtained connectomic gradients for each hemisphere separately (left and right intra-hemispheric) as well as those describing functional connectivity from left to right and right to left hemispheres (left and right inter-hemispheric). We then computed the difference between individual gradient scores to study the asymmetry, consistent with prior studies (Liang et al., 2021; Gonzalez Alam et al., 2022). Subsequently, to evaluate the heritability of possible differences between left and right intra- and inter-hemispheric FC gradients, we used the twin pedigree set-up of the Human Connectome Project S1200 release young adults dataset (Van Essen et al., 2013). To assess whether asymmetry is conserved in other primates, we compared the asymmetry of functional gradients of humans with those observed in macaque monkeys using the prime-DE dataset (Xu et al., 2020; Milham et al., 2018). Finally, we conducted a confirmatory meta-analysis to explore the relationship between the patterns of gradient asymmetry and task-based functional MRI activations. Multiple analyses verified the robustness and replicability of our results.

Results

Hemispheric functional connectivity gradients (Figure 1)

Figure 1 with 2 supplements see all
Processing of functional gradients in humans.

(a) Parcellation using Glasser atlas (Glasser et al., 2016) in each hemisphere and Cole-Anticevic (CA) networks (Ji et al., 2019) for humans. (b) Individual FC in each hemispheric pattern, that is left-left (LL, intra-hemisphere), right-right (RR, intra-hemisphere), left-right (LR, inter-hemisphere), and right-left (RL, inter-hemisphere). (c) Time series of two parcels and the mean functional connectivity (FC) matrix between left and left hemisphere (LL). (d) Gradient template using the group-level gradient of LL. Dots represent parcels and are colored according to CA networks. The decomposition scatter on the right below depicts x-axis (number of eigenvectors) and y-axis (the contribution of each eigenvector to the total). (e) Correlation between left and right mean gradients across subjects of intra- and inter-hemispheric patterns. Left panel is the correlation between gradients of FC LL and FC RR (intra-hemispheric pattern). Right panel is the correlation between gradients of FC LR and FC RL (inter-hemispheric pattern). All correlation coefficients along G1, G2, and G3 are greater than 0.9.

To obtain intra-hemispheric gradients, we first computed the functional connectivity (FC) in 180 homologous parcels per hemisphere using a multimodal parcellation (MMP, Glasser et al., 2016) for each subject (n=1014). For the network level analyses, we employed the Cole-Anticevic atlas (Ji et al., 2019) based on the MMP (Figure 1a). For each individual, FC was summarized in two different patterns (Figure 1b): FC within the left hemisphere (LL mode, intra-hemispheric pattern), within the right hemisphere (RR mode, intra-hemispheric pattern), from left to right hemisphere (LR mode, inter-hemispheric pattern), and from right to left hemisphere (RL mode, inter-hemispheric pattern). We selected the LL mode as the reference template for the gradients approach, and therefore assessed the mean FC that was determined by averaging LL FC across subjects (lower panel in Figure 1c). Here, the reference matches the order and direction of the gradient but does not rescale the gradients. The template gradients were computed by implementing diffusion map embedding, a non-linear dimension reduction technique (Coifman et al., 2005), on the mean LL FC using BrainSpace (Vos de Wael et al., 2020). The current study analyzed asymmetry and its heritability using the first three gradients that explained the most variance (Figure 1d). Each gradient has reasonably well-described functional associations (G1: unimodal-transmodal gradient with 24.1%, G2: somatosensory-visual gradient with 18.4%, G3: multi-demand gradient with 15.1%). However, given that we extracted 10 gradients to maximize the degree of fit (Margulies et al., 2016; Mckeown et al., 2020). We describe mean asymmetry of G4-10 in Figure 1—figure supplement 1.

Next, individual gradients were computed for each subject and the four different FC modes and aligned to the template gradients with Procrustes rotation. It was applied without a scaling factor so that the reference template only matters for matching the order and direction of the gradients. The procedure rotates a matrix to maximum similarity with a target matrix minimizing the sum of squared differences. As noted, Procrustes matching was applied without a scaling factor so that only the reference template matters for matching the order and direction of the gradients. Therefore, it allows comparison between individuals and hemispheres. The individual mean gradients showed high correlation with the group gradients LL (all Spearman r>0.97, P spin <0.001). Figure 1e shows the correlation between LL and RR, LR, and RL modes. In each case, the gradients were highly similar. Similar to previous work (Coifman et al., 2005) we observed that the principal gradient (G1) traversed between unimodal regions and transmodal regions (e.g. default-mode network: DMN) whereas a visual to somatosensory gradient was found for G2. The tertiary gradient (G3) dissociated control from DMN and sensory-motor networks (Figure 1d and e, and Figure 1—figure supplement 2). We employed spin permutations for correcting spatial Spearman correlation p values, that is p spin. For the intra-hemispheric pattern, the mean gradients of LL were strongly correlated with those of RR (Spearman rG1=0.988, Pspin <0.001, rG2=0.989, Pspin <0.001, rG3=0.967, Pspin <0.001). For the inter-hemispheric pattern, the mean gradients of LR were also strongly correlated with those of RL (Spearman rG1=0.993, Pspin <0.001, rG2=0.985, Pspin <0.001, rG3=0.969, Pspin <0.001).

Asymmetry of functional gradients in humans (Figure 2)

Figure 2 with 10 supplements see all
Asymmetry of functional gradients in humans and its heritability.

(a) Mean asymmetry index (AI) of intra- and inter-hemispheric patterns in humans. Red and blue indicate rightward and leftward asymmetry respectively. (b) FDR correction for the P values of AI shown in A; (c) Violin plots of mean AI network loading across individuals (n=1014), with median, 25%-75%, and distribution at 25/75% -/+1.5 interquartile range. Networks are ranked from leftward (language) to rightward asymmetry (frontoparietal) along the intra-hemispheric principal gradient.

Next, we computed the asymmetry index (AI) by subtracting the right hemispheric gradient scores of each parcel from the corresponding left hemispheric scores for our intra- and inter-hemispheric connectivity patterns (Figure 2a). A red AI indicates rightward dominance in gradient scores, whereas blue indicates leftward dominance. The differences in gradient loadings (parcel No.25: Peri-Sylvian language area) reflect differences in connectivity profiles (top 10%) between LL versus RR, or LR versus RL, respectively (Figure 2—figure supplement 1). The significance of AI scores for the intra- and inter-hemispheric patterns were reported after false discovery rate adjustment (P FDR < 0.05) (Figure 2b), and Cohen’s d maps can be seen in Figure 2—figure supplement 2. Frontal and temporal lobes showed the greatest intra-hemispheric asymmetry in G1 (Supplementary file 1). In particular, regions in ventral- and dorsolateral PFC (11 l, p9-46v, p10p) were the three most rightward asymmetric areas and regions in temporal polar cortex, dorso/posterior superior temporal sulcus, and inferior frontal gyrus (TGv, STSdp, and 55b) were the three most leftward asymmetric areas in the intra-hemispheric pattern. Network-level analyses (Figure 2c) indicated that the language (t=41.3, df = 1013, P FDR < 0.001) and default mode (t=17.3, df = 1013, P FDR < 0.001) networks had a high leftward AI, while the frontoparietal network (t=–26.0, df = 1013, P FDR < 0.001) had a high rightward AI. We observed no significant difference of AI in primary and secondary visual networks. Overall, asymmetry was widely present along the first three connectivity gradients, including G2 and G3. Detailed numbers can be seen at online ipython notebook (code availability).

For the inter-hemispheric pattern, a large portion of the cerebral cortex showed significant AI scores. The top six asymmetric areas included regions in inferior frontal cortex and parahippocampal regions (11 l, 47 m, p9-46v, PSL, PreS, and PHA2) (Supplementary file 1). At the network level (Figure 2c), networks with leftward dominance were the visual (tprimary visual = 9.3, df = 1013, P FDR < 0.001; t secondary visual = 7.5, df = 1013, P FDR < 0.001), language (t=5.7, df = 1013, P FDR < 0.001), default mode (t=11.9, df = 1013, P FDR < 0.001), and orbito-affective (t=4.6, df = 1013, P FDR < 0.001) networks. Networks with rightward dominance were the somatomotor (t=–3.5, df = 1013, P FDR = 0.00059), cingulo-opercular (t=–14.6, df = 1013, P FDR < 0.001), dorsal attention (t=–8.0, df = 1013, P FDR < 0.001), frontoparietal (t=–12.1, df = 1013, P FDR < 0.001), and auditory (t=5.7, df = 1013, P FDR < 0.001) networks. Posterior and ventral multimodal networks were not significantly asymmetric.

The mean AI scores across individuals for the intra- and inter-hemispheric patterns showed high similarity (Spearman r G1 = 0.645, P spin <0.001). This may indicate that the asymmetric functional organization is a feature that is captured both by inter- and intra-hemispheric connectivity patterns.

Heritability of asymmetry of functional gradients in humans (Figure 3)

Figure 3 with 1 supplement see all
Heritability of asymmetry of functional G1.

(a) Heritability (orange colorbar) and p values after FDR correction (green colorbar). (b) Scatter plot of heritability and AI scores. The x- and y-axes are the mean asymmetry index and heritability, respectively. Dots represent parcels and are colored according to CA networks. The small scatter plots with a regression line are the corresponding absolute mean asymmetry index (x-axis) and heritability (y-axis).

We next computed the heritability of the AI scores of the functional gradient for the intra- and inter-hemispheric patterns using Solar-Eclipse 8.5.1 beta (http://solar-eclipse-genetics.org/). We found that left-right differences observed in large-scale functional organization axes were heritable (Figure 3a). Specifically, for the intra-hemispheric pattern, we found sensory-motor regions, middle temporal regions, dorso-lateral, and medial prefrontal regions to be heritable (P FDR < 0.05). In the case of the inter-hemispheric pattern, all cortical regions with the exception of visual areas and superior temporal and insular regions were heritable (P FDR < 0.05). Notably, language-associated areas such as the PSL (Peri-Sylvian language area) and 55b had the highest heritability in both the hemispheric patterns (PSL: intra: h2=0.46, P FDR < 0.001 and inter: h2=0.34, P FDR < 0.001, Supplementary file 1). However, BA area 44 (Broca’s area) showed low heritability (intra: hGüntürkün et al., 2020 = 0.12, P FDR = 0.026 and inter: h2=0.12, P FDR = 0.018). The G2 and G3 results are shown in Figure 3—figure supplement 1.

To assess whether regions showing higher asymmetry had an increased heritability of G1, we plotted our cortical maps of asymmetry along those reporting heritability (Figure 3b). For the correlation between the absolute asymmetry index and heritability (Figure 3b small scatter), gradients of the intra-hemispheric FC patterns were significant (Pearson r=0.245, P spin = 0.005) while gradients of the inter-hemispheric FC were not (Pearson r=0.055, P spin = 0.613).

Asymmetry of functional gradients in macaques (Figure 4)

Figure 4 with 1 supplement see all
Asymmetry of functional gradients in macaques.

(a) Parcellation used Markov atlas in macaques Markov et al., 2014. (b) Template gradients of group level connectivity of LL. (c) Mean asymmetry index of G1 in macaques. (d) Normalized (Cohen’s d) asymmetry of G1 in macaques and humans aligned to macaque’s surface. Purple indicates leftward asymmetry, whereas yellow indicates rightward asymmetry. (e) Similarity of normalized asymmetry of G1 between humans and macaques. (f) The details of how the human Cole-Anticevic network atlas is projected to the macaque surface can be seen in the Methods. Bold colors indicate human mean cohen's D values in a given network and pastel colors indicate macaque mean cohen's D values in a given network. Networks are ranked from leftward (language) to rightward asymmetry (frontoparietal) along the intra-hemispheric principal gradient in humans for comparison.

To probe the phylogenetic conservation of asymmetry of functional organization in primates, we performed the same diffusion map embedding analysis on macaque resting-state FC data (n=19, PRIMATE-DE sample Xu et al., 2020; Milham et al., 2018). We used the Markov parcellation (Markov et al., 2014) in macaques, resulting in 91 parcels per hemisphere (Figure 4a) and then computed FC in the four patterns: LL and RR (intra-hemispheric patterns), and LR and RL (inter-hemispheric patterns). Following the same connectome gradients analysis pipeline as deployed on the human FC data, we obtained the template gradients on the LL intra-hemispheric FC pattern (Figure 4b). The first three template gradients explained 20.0%, 15.2%, and 12.8% of total variance, respectively. G1 described an axis traversing dorsolateral prefrontal and parietal regions (anterior-posterior).

Evaluating the intra-hemispheric pattern of functional organization in macaques along G1, we observed that parietal cortices had a rightward dominance while occipital cortices were leftward. Temporal cortex asymmetry was low (Figure 4c). The inter-hemispheric pattern showed similar asymmetry to the intra-hemispheric pattern along G1. However, the AI scores of the principal, but also secondary and tertiary gradients, were not statistically significant after FDR correction, both for intra- and inter-hemispheric patterns. The effect sizes across cortex observed in macaques along G1 were [intra: mean Cohen’s d=–0.27 (rightward) and 0.27 (leftward); inter: mean Cohen’s d=–0.22 (rightward) and 0.20 (leftward)].

To compare human and macaque connectomic gradients, we aligned human gradients to the same macaque surface space (Figure 4d) using a joint embedding technique (Xu et al., 2020). We summarized Cohens’ d of AI of macaque-aligned human gradients within the Markov parcels for the intra- and inter-hemispheric patterns and compared the similarity of Cohens’ d of AI between the two species using Spearman correlations (Figure 4e). To reduce the systematic bias during the cross-species alignment, we averaged the results of left and right hemispheric alignment. We found that the macaque and macaque-aligned human AI maps of G1 were correlated positively for intra-hemispheric patterns (Pearson r=0.345, P spin = 0.030). For inter-hemispheric patterns, we did not observe a significant association (Pearson r=–0.029, P spin = 0.858).

We then projected the human functional networks (Ji et al., 2019) on the macaque surface (Xu et al., 2020), to qualitatively compare differences in human functional networks between humans and macaques (Figure 4f and Figure 4—figure supplement 1). In the case of the intra-hemispheric asymmetry of the principal FC gradient, we observed that humans showed high leftward asymmetry in the language and default mode networks but macaques did not. Moreover, humans showed high rightward asymmetry in the frontoparietal and cingulo-opercular networks but macaques did not. Humans and macaques showed an opposite direction of asymmetry in auditory, orbito-affective, and secondary visual networks. For the inter-hemispheric FC pattern, macaques and humans showed only subtle differences.

Functional decoding along the normalized asymmetry of G1 (Figure 5)

Figure 5 with 1 supplement see all
Projection of meta-analytical task-based function along normalized asymmetry of G1 (intra-hemisphere).

The 20 bins were generated by normalized (Cohen’s d) asymmetry of G1 in humans. Cool color indicates regions showing leftward dominance and warm color indicates regions showing rightward dominance. The order of the terms of the y-axis was generated by the weighted score of activation (z-score >0.5) * normalized asymmetry.

Finally, we investigated the relationship between patterns of asymmetry of functional organization in humans and task-based meta-analytic functional activations. To do so, we projected meta-analytical fMRI activation maps (Yarkoni et al., 2011) along the normalized (Cohen’s d) asymmetry of G1 (Figure 5). Our choice for the 24 cognitive domain terms were consistent with prior literature (Margulies et al., 2016). Here, we calculated the weighted score by activation z-score (parcels where activation z-score was greater than 0.5) multiplied by the normalized asymmetry, suggesting leftward to rightward preference, seen from top to bottom of the y-axis of Figure 5. Language, semantics, and reading domains were associated with leftward hemispheric preference, whereas cognitive control, inhibition, and working memory were associated with rightward hemispheric preference. For the asymmetry of the inter-hemispheric FC gradient, we observed a similar pattern of association (Figure 5—figure supplement 1). This indicates that patterns of asymmetry in functional organization also align with task-based activations consistently reported in the literature.

Robustness analyses

Complementing our main AI calculation (L-R), we additionally used AI_norm (L-R)/(L+R), with rescaling the distribution of gradients to positive values, to explore whether our results were robust with respect to AI calculation (Figure 2—figure supplement 3). We found that for G1, asymmetric effects were highly correlated with the main asymmetric effects (Spearman r intra- hemisphere = 0.851, P uncorrected <0.001; r inter-hemisphere = 0.863, P uncorrected <0.001). Significant correlation was also found in G2 (Spearman r intra-hemisphere = 0.681, P uncorrected <0.001; r inter-hemisphere = 0.228, P uncorrected = 0.002) and in G3 (Spearman r intra-hemisphere = 0.795, P uncorrected <0.001; r inter- hemisphere = 0.879, P uncorrected <0.001).

To test the robustness of our findings with respect to the parcellation approach, we employed Desikan-Killiany atlas 1 to generate the asymmetry of functional gradients. This is a symmetric atlas containing 34 parcels per hemisphere. Overall, for the intra-hemispheric pattern G1 showed similar hemispheric patterns as observed in our main results when using the Desikan-Killiany atlas. In particular, the posterior cluster between middle and superior temporal gyrus and Broca’s area showed leftward asymmetry, whereas dorsolateral prefrontal regions showed rightward asymmetry (Figure 2—figure supplement 4). However, we observed more details are shown in multi-modal parcellation. Similar patterns were observed in inter-hemispheric asymmetry of functional organization when using the Desikan-Killiany atlas.

We also used an additional sample (UK Biobank, UKB) to verify whether asymmetry of functional organization is present in other samples. We included 34,830 subjects’ imaging data in UKB with good quality. After computing Cohen’s d of asymmetric effects in UKB, to account for differences in sample size, we performed a group level correlation between HCP with UKB (Figure 2—figure supplement 5). We observed a high correlation between the LL functional gradient between HCP and UKB (Spearman r G1 = -0.588, P uncorrected <0.001; Spearman r G2 = 0.309, P uncorrected <0.001; Spearman r G3 = 0.773, P uncorrected <0.001). Thus, we flipped UKB LL G1 direction to make it more consistent with HCP (now r G1 = 0.588, P uncorrected <0.001). For the G1 of the intra-hemispheric FC pattern, we observed a correlation with our findings in the HCP sample (Pearson r intra-hemisphere = 0.592, P spin <0.001; Pearson r inter-hemisphere = 0.384, P spin <0.001). All the networks showed significant asymmetry in UKB. However, we found that language and default mode networks showed leftward asymmetry (as in HCP), but the frontoparietal network did not show rightward asymmetry.

Evaluating possible effects due to the parcellation scheme used, we studied differences of the mean rsfMRI connectome along the first gradient at the vertex level. We used 100 random subjects, as we had the data mapped to a symmetric template (fs_LR_32 k), which indicated that each vertex has a symmetric counterpart in the right hemisphere. Our results show left-right asymmetry being language/default mode-visual-frontoparietal vertices, which is consistent with the main results of the parcel-based approach (Figure 2—figure supplement 6).

To evaluate potential downstream effects of alignment to our results, we compared the gradient asymmetry with Procrustes alignment to the gradient without alignment. This resulted in virtually identical results for the HCP sample (r intra-hemisphere = 0.956, r inter-hemisphere = 0.843, Figure 2—figure supplement 7). At the same time, comparing unaligned and aligned gradients in the UKB sample, we found that the alignment improved the similarity to the pattern observed in HCP (aligned r intra-hemisphere = 0.592, non-aligned r intra-hemisphere = 0.487, aligned r inter-hemisphere = 0.384, non-aligned r inter-hemisphere = 0.162, Figure 2—figure supplement 8).

Moreover, to overcome potential normalization biases associated with creating one gradient for each hemisphere, we performed an alternative analysis to create a gradient of the left and right hemisphere together. This assumes that regions with similar connectivity profiles have comparable loading in the gradient framework. Indeed, along the principal gradient, the observed normalized asymmetric map was highly similar to the non-normalized map used in the main analyses for the intra-hemispheric (Pearson r=0.956) and inter-hemispheric (Pearson r=0.531) asymmetry patterns (Figure 2—figure supplement 9). It is possible the difference between intra- and inter-hemispheric correspondence relates to more global differences in strength of connectivity comparing LR to RL FC, as reported also in the article (Raemaekers et al., 2018) resulting in more widespread differences between inter-hemispheric patterns of both embedding procedures.

Finally, we also set the RR FC gradients as reference for our analyses, the first three of which explained 22.8, 18.8, and 15.9% of the total variance. We aligned each individual to this reference (Figure 2—figure supplement 10). It suggested all results were virtually identical (Pearson r intra G1=0.989, r intra G2=0.939, r intra G3=0.987, r inter G1=0.979, r inter G2=0.960, r inter G3=0.990, all P spin <0.001).

Discussion

In this study, we investigated the extent to which human cortical functional organization is asymmetric using a gradient-based approach. We assessed whether genetic factors shape such asymmetry and evaluated whether patterns of asymmetry are phylogenetically conserved between humans and non-human primates (macaques). We found that the principal gradient revealed hemispheric differences in most cortical regions, excluding the visual cortex. The language network and default-mode network showed the most leftward asymmetry while the frontoparietal network showed the most rightward asymmetry. The observed asymmetry of functional organization along the principal gradient was heritable. At the same time, regions with high asymmetry showed variable heritability. This may suggest that asymmetry in functional organization reflects both heritable and experience-dependent factors. Although the difference in left and right hemispheric functional organization was not significant along the principal functional gradient in a sample of macaques, the inter-hemispheric asymmetric pattern was comparable to the asymmetry pattern observed in humans indicating phylogenetic conservation. Notably, both the language and frontoparietal networks showed a higher leftward asymmetry in humans relative to macaques, indicating cross-species differences in asymmetry of specific transmodal functional networks. Decoding task-based functional activations along the asymmetry axis of the principal gradient, we observed that regions with a leftward preference were associated with language, autobiographical memory, and social cognition domains, whereas those with a rightward preference included cognitive control, working memory, and inhibition. In sum, our study shows the asymmetry of functional organization is, in part, heritable in humans and phylogenetically conserved in humans and macaques. At the same time, we observed that asymmetry of regions linked to higher-order cognitive functions such as language and cognitive control showed marked differences between humans and macaques and variable heritability in humans, possibly reflecting an evolutionary adaptation allowing for experience-dependent specialization.

By studying asymmetry in functional organization using a gradient approach, we have extended previous studies reporting asymmetric functional connectivity. Indeed, although the functional organization of the cerebral cortex has a largely symmetric pattern, it also shows subtle differences between hemispheres (Liang et al., 2021; Gonzalez Alam et al., 2022; Iturria-Medina et al., 2011; Sun et al., 2017). For the intra-hemispheric asymmetry gradients, we found that regions belonging to the language network showed the strongest leftward preference along the principal gradient axis. This indicates that their functional connectivity profiles were more similar to the default mode, relative to their right-hemispheric counterparts. Conversely, ventral multimodal networks were closer to the transmodal apex of the principal gradient in the right hemisphere, relative to their homologues in the left hemisphere. As such, our observations suggest that key transmodal regions, part of the language and control networks, show organizational preference to either the left or right hemisphere. Anterior lateral default mode subnetworks have been shown to uniquely exhibit positive connectivity to the language network (Gordon et al., 2020), possibly leading to increased gradient loadings of the language network in the left hemisphere, placing them closer to the default regions along the principal gradient in the left hemisphere relative to the right. Conversely, the transmodal frontoparietal network was located at the apex of rightward preference, possibly suggesting a rightward lateralization of cortical regions associated with attention and control and ‘default’ internal cognition (Corbetta and Shulman, 2002; Smallwood et al., 2021). The observed dissociation between language and control networks is also in line with previous work suggesting an inverse pattern of language and attention between hemispheres (Karolis et al., 2019; Zago et al., 2016). Such patterns may be linked to inhibition of corpus callosum (Cook, 1984), promoting hemispheric specialization. It has been suggested that such inter-hemispheric connections set the stage for intra-hemispheric patterns related to association fibers (Gazzaniga, 2000). Future research may relate functional asymmetry directly to asymmetry in underlying structure to uncover how different white-matter tracts contribute to asymmetry of functional organization.

We furthermore investigated whether such individual variations in asymmetry of functional organization could be attributed to genetic factors. To do so, we performed heritability analysis enabled by the twin design of the Human Connectome Project (Van Essen et al., 2013). Previous work indicated that brain structure including cortical thickness, surface area, and white matter connection (Kong et al., 2018; Sha et al., 2021; Zhong et al., 2021), as well as functional connectome organization (Colclough et al., 2017; Ge et al., 2017) are heritable. Our twin-based heritability analyses revealed heritable asymmetry of the principal functional gradient in the entire cortex, excluding visual cortex. At the same time, studying the association between heritability and asymmetry patterns we observed mixed results. Although we observed that the language-related area PSL showed the highest heritability, the highly asymmetric area 44 (Broca’s area) showed the lowest heritability. This may reflect a differential (dorsal and ventral) pathway of language development in the frontal and temporal lobe, where the dorsal pathway to the inferior frontal gyrus matures at later stages in development (Brauer et al., 2013). For example, previous work found that temporal language areas showed high heritability of cortical thickness asymmetry (Sha et al., 2021) and white matter connection asymmetry (Zhong et al., 2021) but frontal language areas did not. Such posterior-anterior differences may be due to developmental factors or axes of stability versus plasticity in the cortex (García-Cabezas et al., 2017). A case study of an individual born without a left temporal lobe found that frontal language areas in the left hemisphere did not emerge in the absence of temporal language areas in the left hemisphere, and that language functions instead relied on the right hemispheric functional network (Tuckute et al., 2022). It is thus possible that Broca’s area may mature after more posterior language regions in hierarchical fashion, which may be related to decreasing heritability in frontal language areas (i.e. more influenced by developmental and/or environmental factors). Recent work suggests that asymmetric patterning of brain structure and function are largely determined prenatally and unaffected by preterm birth (Williams et al., 2021). In neonates, asymmetric patterns were largely observed in primary and unimodal areas, whereas association regions were largely symmetric. Thus, asymmetry in association regions may be more experience-dependent. One focus of future work could thus be to evaluate the development of asymmetry in functional organization. Moreover, by means of GWAS approaches, it may also be possible to get more insight in specific genes and associated processes involved in functional asymmetry.

Evaluating the correspondence of asymmetry of functional organization between humans and macaques, by aligning the human gradients to the macaque gradient space (Xu et al., 2020), we observed a similarity between asymmetry of the principal gradient in both species in case of intra-hemispheric connectivity. This indicates functional asymmetry of within-hemispheric connectivity may be conserved across primates. At the same time, we found that language, default mode, and frontoparietal networks showed qualitatively more asymmetry in humans (human >macaque). These findings may support the notion that though asymmetry is a phenomenon existent across different primates, regions involved in higher order cognitive functions in humans are particularly asymmetric. Previous work studying asymmetry in white matter structure in primates found that humans showed more leftward arcuate fasciculus volume and surface relative to macaques (Eichert et al., 2019). The arcuate fasciculus is a white matter tract implicated in language functions by connecting Broca’s and Wernicke’s areas (Geschwind, 1970). Moreover, by comparing humans, macaques, and chimpanzees (Eichert et al., 2020), evolutionary modifications to this tract in humans relative to other primates have been reported, possibly derived from auditory pathways (Balezeau et al., 2020). At the same time, other structural studies have also observed leftward asymmetry of language areas in chimpanzees, indicating that asymmetry of language-regions per se may not be a human-specific feature (Spocter et al., 2020; Xiang et al., 2020). Fittingly, there are no significant differences of thickness and area asymmetry between humans and chimpanzees in superior temporal lobe (Xiang et al., 2020). Studying the endocranial shape of humans and non-human primates, temporal and occipital cortices showed local differences in asymmetry across species, and much more variability in humans relative to non-human primates (Neubauer et al., 2020). This suggests that whereas brain asymmetry is a phenomenon observed throughout mammals (Assaf et al., 2020), specific nuances may relate to species-specific behavioral and cognitive differences. Future research could assess asymmetry of brain organization in other primates, and relate observed differences in functional organization to those in white matter structure.

Although overall intra- and inter-hemispheric connectivity showed a strong spatial overlap in humans, we also observed various differences between the metrics across our analysis. For example, although we found both intra- and inter-hemispheric differences in gradient organization to be heritable, only for the former was a correspondence between the degree of asymmetry and heritability found. Similarly, comparing human and macaques, we only observed conservation of spatial patterning of asymmetry was conserved for intra-hemispheric connections. Whereas intra-hemispheric asymmetry relates to association fibers, commissural fibers underlie inter-hemispheric connections (Tzourio-Mazoyer, 2016). It has been suggested that there is a trade-off within and across mammals of inter- and intra-hemispheric connectivity patterns to conserve the balance between gray and white-matter (Assaf et al., 2020). Consequently, differences in asymmetry of both ipsi- and contralateral functional connections may be reflective of adjustments in this balance within and across species. Secondly, previous research studying intra- and inter-hemispheric connectivity and associated asymmetry, has indicated a developmental trajectory from inter- to intra-hemispheric organization of functional brain connectivity, varying from unimodal to transmodal areas (Friederici et al., 2011; Szaflarski et al., 2006). It is thus possible that a reduced correspondence of asymmetry and heritability in humans, as well as lack of spatial similarities between humans and macaques for inter-hemispheric connectivity may be due to the age of both samples (young adults in humans, adolescents in macaques). Further research could study inter- and intra-hemispheric asymmetry in functional organization as a function of development in both species, to further disentangle heritability and cross-species conservation and adaptation.

The functional relevance of asymmetry along the sensory-transmodal axis was evaluated in the human brain by projecting meta-analytical task-based coactivations along asymmetric effects of the functional principal gradient. In line with our expectations based on the distribution of asymmetry within functional networks, we found that task-based activations associated with language processing leaned leftward while task-based activations associated with executive functions leaned rightward, specifically in the intra-hemispheric pattern. This suggests that lateralized functions supported by the brain’s asymmetry have functional relevance (especially higher order cognitive functions such as language and executive control). Indeed, related work has shown a direct link between asymmetry and semantic and visual recognition skills (Gonzalez Alam et al., 2022), suggesting that asymmetry of individuals relates to variation in behavioral performance in these domains (Gonzalez Alam et al., 2022; Gonzalez Alam et al., 2019). Our observation of asymmetry of language versus executive functions may also be in line with notions of differential axes of asymmetry, dissociating symbolic/language, emotion, perception/action, and decision functional axes (Karolis et al., 2019). The asymmetry of principal functional gradient in humans and macaques showed a divergence along these axes, possibly indicating cross-species variability within the lateralization archetypes in primates. Notably, left hemispheric language lateralization is enabled throughout language development while right hemispheric language activation declines systematically with age (Olulade et al., 2020). Therefore, future research may focus on studying how the lateralization of human behavior is shaped by development and aging and how this may impact function and behavior.

Although we showed asymmetry in functional organization, there are various technical and methodological aspects to be considered. In the current work, we used the MMP (Glasser et al., 2016) for surface-based human fMRI data. A previous study used the atlas of intrinsic connectivity of homotopic areas (Xiang et al., 2020) (AICHA, https://www.gin.cnrs.fr/en/tools/aicha) for voxel-based fMRI data (Liang et al., 2021). In line with the results of that study, we found similar intra-hemispheric differences in functional gradients. Extending that work we additionally used the DK atlas (Desikan et al., 2006), which is often used in structural asymmetry studies (Sha et al., 2021; Postema et al., 2019). We again found asymmetric patterns, with a rightward dorsal frontal lobe and leftward posterior superior temporal lobe. The other temporal regions, having leftward or rightward asymmetry using MMP, showed no or less asymmetry using the DK atlas. Possibly, such subtle differences are not captured by the DK atlas, with only 34 cortical parcels per hemisphere. Evaluating the consistency of functional asymmetry across different datasets, we found that HCP (n=1014) and UKB (n=34,604) showed consistent leftward asymmetric functional organization in the language and default mode networks but no consistent rightward asymmetry of the frontoparietal network. Such differences may be due to technical differences between the datasets (Xifra-Porxas et al., 2021). However, it may also reflect sample specific differences in asymmetry. Indeed, whereas the HCP sample consists of young-adults with an age-range of 22–37 years, the UKB has a comparatively older and wider age range (from 40 to more than 70 years). Thus, it is possible the observed differences in the frontoparietal network are directly related to age-related asymmetry effects (Olulade et al., 2020). Due to the small sample size of macaques, it is important to be careful when interpreting our observations regarding the associated asymmetry, and its relation to patterns observed in humans. Therefore, further study is needed to evaluate the asymmetry patterns in macaques using large datasets (Milham et al., 2018; Messinger et al., 2021).

To conclude, we investigated the genetic and phylogenetic basis of asymmetry of large-scale functional organization. We observed that the principal (unimodal-transmodal) gradient (Margulies et al., 2016) is asymmetric, with regions involved in language showing leftward organization and regions associated with executive function showing rightward organization. This asymmetry was heritable and, in the case of organization of intra-hemispheric connectivity, showed spatial correspondence between humans and macaques. At the same time, functional asymmetry was more pronounced in language networks in humans relative to macaques, suggesting adaptation. The current framework may be expanded by future research investigating the development and phylogeny of functional asymmetry as well as its neuroanatomical basis in healthy and clinical samples. This may provide important insights in individual-level brain asymmetry and its relation to human cognition.

Materials and methods

The current research complies with all relevant ethical regulations as set by The Independent Research Ethics Committee at the Medical Faculty of the Heinrich-Heine-University of Duesseldorf (study number 2018–317).

Participants

Humans

Request a detailed protocol

For the analyses in humans, we used the Human Connectome Project (HCP) S1200 data release (Van Essen et al., 2013). That release contains four sessions of resting state (rs) fMRI scans for 1206 healthy young adults and their pedigree information (298 monozygotic and 188 dizygotic twins as well as 720 singletons). We included individuals with a complete set of four fMRI scans that passed the HCP quality assessment (Van Essen et al., 2013; Glasser et al., 2013). Finally, our sample consisted of 1014 subjects (470 males) with a mean age of 28.7 years (range: 22–37).

For the replication, we employed the UKB dataset (application ID: 41655) including 34,830 subjects’ imaging data. Details on data processing and acquisition can be found in the UKB Brain imaging documentation (https://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/brain_mri.pdf). Briefly, resting-state imaging data was motion corrected, intensity normalized, high-pass temporally filtered, and further denoised using the ICA-FIX pipeline, all implemented in FSL. MPM parcellation was warped to subject-space based on the high-resolution T1-weighted anatomical image. Individual warping parameters were applied to map the MPM parcellation to the functional space following T1-rsfMRI alignment. The age range of the UKB sample was from 40 to more than 70 years.

Macaques

Request a detailed protocol

We selected rhesus macaque monkeys’ rs-fMRI data from the non-human primate (NHP) consortium PRIME-DE (http://fcon_1000.projects.nitrc.org/indi/indiPRIME.html) from Oxford. The full dataset consisted of 20 rhesus macaque monkeys (macaca mulatta) scanned on a 3T with a 4-channel coil (Noonan et al., 2014). The rs-fMRI data were collected with 2 mm isotropic resolution, TR = 2 s, 53.3 mins (1600 volumes). Details can be seen in Xu et al., 2020. Nineteen macaques with successful preprocessing and surface reconstruction were included in the current study (all males, age = 4.01 ± 0.98 years, weight = 6.61 ± 2.04 kilograms).

Macaque data were preprocessed with an HCP-like pipeline (Xu et al., 2015)⁠, described elsewhere (Xu et al., 2020). In brief, it included temporal compression, motion correction, 4D global scaling, nuisance regression using white matter (WM), cerebrospinal fluid (CSF), and Friston-24 parameter models, bandpass filtering (0.01–0.1 Hz), detrending, and co-registration to the native anatomical space. The data were then projected to the native midcortical surface and smoothed along the surface with FWHM = 3 mm. Finally, the preprocessed data were down-sampled to the surface space (with resolution of 10,242 vertices in each hemisphere).

Parcellations

Multimodal parcellation and Cole-Anticevic network

Request a detailed protocol

We used multimodal parcellation (MMP) of 360 areas (180 per hemisphere) for humans (Glasser et al., 2016)⁠. This atlas has been generated using the gradient-based parcellation approach with similar gradient ridges presenting in roughly corresponding locations in both hemispheres, which is suitable for studying asymmetry across homologous parcels. Additionally, based on MMP, we used the Cole-Anticevic Brain-wide Network Partition (CA network), which includes in total 12 functional networks (Ji et al., 2019).

Desikan-Killiany atlas

Request a detailed protocol

To ensure our results were reliable we repeated the analysis in humans using a different brain atlas. The Desikan-Killiany atlas (Desikan et al., 2006) contains 34 cortical parcels per hemisphere in humans and has high correspondence across two hemispheres.

Markov parcellation

Request a detailed protocol

For the macaques, we used 91 cortical areas per hemisphere in the Markov M132 architectonic parcellation (Markov et al., 2014). This directed and weighted atlas is generated based on the connectivity profiles. The 91-area parcellation in macaques is valuable for comparison with connectivity analyses in humans.

Functional connectivity

Request a detailed protocol

All rs-fMRI data underwent HCP’s minimal preprocessing (Glasser et al., 2013) and were coregistered using a multimodal surface matching algorithm (MSMAll) (Robinson et al., 2014) to the HCP template 32 k_LR surface space. The template consists of 32,492 total vertices per hemisphere (59,412 excluding the medial wall). Cortical time series were averaged within a previously established multi-modal parcellation schemes: for humans the 360-parcel Glasser atlas (180 per hemisphere) (Glasser et al., 2016) and the 182-parcel Markov atlas (91 per hemisphere) for macaques (Markov et al., 2014). To compute the functional connectivity (FC), time-series of cortical parcels were correlated pairwise using the Pearson product moment and then Fisher’s z-transformed in human and macaque data, separately. Individual FC maps were also averaged across four different rs-fMRI sessions for humans ([LR1], [LR2], [RL1], and [RL2]). We computed the FC in four different patterns, both for human and macaque data: FC within the left and right hemispheres (LL intra-hemisphere, RR intra-hemisphere), from the left to right hemisphere (LR inter-hemisphere) and from the right to left hemisphere (RL, inter-hemisphere).

Connectivity gradients

Request a detailed protocol

Next we employed the nonlinear dimensionality reduction technique (Margulies et al., 2016) to generate the group level gradients of the mean LL FC across individuals. We then set the group-level gradients as the template and aligned each individual gradient with Procrustes rotation to the template. Finally, the comparative individual functional gradients of each FC pattern were assessed. All steps were accomplished in the Python package Brainspace (Vos de Wael et al., 2020). In brief, the algorithm estimates a low-dimensional embedding from a high-dimensional affinity matrix. Along these low-dimensional axes, or gradients, cortical nodes that are strongly interconnected, by either many suprathreshold edges or few very strong edges, are closer together. Nodes with little connectivity similarly are farther apart. Regions having similar connectivity profiles are embedded together along the gradient axis. The name of this approach, which belongs to the family of graph Laplacians, is derived from the equivalence of the Euclidean distance between points in the diffusion embedded mapping (Coifman et al., 2005; Margulies et al., 2016; Vos de Wael et al., 2020). It is controlled by a single parameter α, which controls the influence of the density of sampling points on the manifold (α=0, maximal influence; α=1, no influence). On the basis of the previous work (Margulies et al., 2016), we followed recommendations and set α=0.5, a choice that retains the global relations between data points in the embedded space and has been suggested to be relatively robust to noise in the covariance matrix.

The input of the analysis was the FC matrix, which was cut off at 90% similar to previous studies (Margulies et al., 2016). The current study selected the first three FC LL gradients (G1, G2, and G3) that explained 24.1, 18.4, and 15.1% of total variance in humans, as well as 18.9, 15.2, and 12.8% of total variance in macaques.

Asymmetry index

Request a detailed protocol

To quantify the left and right hemisphere differences, we chose left-right as the asymmetry index (AI) (Liang et al., 2021; Raemaekers et al., 2018). In addition, we also calculated the normalized AI with the following formula, (left-right)/(left +right), which is usually used in structural studies to verify whether there is a difference between unnormalized AI and normalized AI. For the intra-hemispheric pattern, the AI was calculated using LL-RR. A positive AI-score meant that the hemispheric feature dominated leftwards, while a negative AI-score dominated rightwards. For the inter-hemispheric pattern we used LR-RL to calculate the AI. Notably, we added ‘minus’ to the AI scores or Cohen’s d scores in the figures in order to conveniently view the lateralization direction.

Heritability analysis

Request a detailed protocol

To map the heritability of functional gradient asymmetry in humans, we used the Sequential Oligogenic Linkage Analysis Routines (SOLAR, v8.5.1b) (Almasy and Blangero, 1998). In brief, heritability indicates the impact of genetic relatedness on a phenotype of interest. SOLAR uses maximum likelihood variance decomposition methods to determine the relative importance of familial and environmental influences on a phenotype by modeling the covariance among family members as a function of genetic proximity (Valk et al., 2020; Almasy and Blangero, 1998). Heritability (i.e. narrow-sense heritability h2) represents the proportion of the phenotypic variance (σ2p) accounted for by the total additive genetic variance (σ2g), that is h2 = σ2g / σ2p. Phenotypes exhibiting stronger covariances between genetically more similar individuals than between genetically less similar individuals have higher heritability. In this study, we quantified the heritability of asymmetry of functional gradients. We added covariates to our models including age, sex, age2, and age ×sex.

Alignment of humans to macaques

Request a detailed protocol

To phylogenetically map the asymmetry of functional gradients across macaques and humans, we transformed the human gradients to macaque cortex surface based on a functional joint alignment technique (Xu et al., 2020). This method leverages advances in representing functional organization in high-dimensional common space and provides a transformation between human and macaque cortices, also previously used in Valk et al., 2020; Valk et al., 2022.

In the present study, we aligned Cohen’s d of the human asymmetry index to the macaque surface. Cohens’ d explains the effect size of the asymmetry index. Following the joint alignment, we further computed the Spearman correlation between macaques and humans to evaluate the similarity in asymmetric patterns of the functional gradients. Finally, we compared Cohen’s d between macaques and humans and summarized the results with Markov parcellation (Markov et al., 2014). To illustrate our findings at the functional network level, we projected human networks (Ji et al., 2019) on the macaque surface.

NeuroSynth meta-analysis

Request a detailed protocol

To evaluate the association of function decoding and asymmetry of the principal gradient, we projected the meta-analytical task-based activation along the normalized asymmetry (Cohen’s d) of G1. Our choice for the 24 cognitive domain terms were consistent with (Margulies et al., 2016). The activation database we used for meta-analyses was the Neurosynth V3 database (Yarkoni et al., 2011). The surface-based V3 database is available in the github depository (data availability). In the present study, to look at how the right hemisphere and left hemisphere decode functions separately, the leftward normalized asymmetry was put on and the rightward normalized asymmetry was put on the right hemisphere. Other regions became zero. We generated 20 bins along the normalized asymmetry averagely (5% per bin). Thus, each function term had a mean activation z-score per bin. To assess how much the function term was leftward or rightward lateralized, we calculated a weighted score by mean activation (where activation z-score greater than 0.5) multiplied by normalized asymmetry. We roughly regarded this score as the lateralization level. The order of the function terms generated by this calculation reflected the left-right lateralization dominance axis.

Data availability

Request a detailed protocol

All human data analyzed in this manuscript were obtained from the open-access HCP young adult sample (https://www.humanconnectome.org/), UK Biobank (https://www.ukbiobank.ac.uk/). Macaque data came from PRIME-DE (http://fcon_1000.projects.nitrc.org/indi/indiPRIME.html). Gradient analyses and visualization were performed using the Python package Brainspace (Vos de Wael et al., 2020) (https://brainspace.readthedocs.io/en/latest/index.html). Heritability analyses were performed using Solar Eclipse 8.5.1b (https://www.solar-eclipse-genetics.org). Task-based function association analyses were based on NeuroSynth (Yarkoni et al., 2011) (https://neurosynth.org/). Full statistical scripts can be found at https://github.com/CNG-LAB/cngopen/tree/main/asymmetry_functional_gradients (copy archived at swh:1:rev:07d4a1a03267dac12ac8bfbccc8e09049cac9f31;path=/asymmetry_functional_gradients; Bayrak et al., 2022).

Data availability

All human data analyzed in this manuscript were obtained from the open-access HCP young adult sample (https://www.humanconnectome.org/), UK Biobank (https://www.ukbiobank.ac.uk/). Macaque data came from PRIME-DE (http://fcon_1000.projects.nitrc.org/indi/indiPRIME.html). Gradient analyses and visualization were performed using the Python package Brainspace (Vos de Wael et al., 2020) (https://brainspace.readthedocs.io/en/latest/index.html). Heritability analyses were performed using Solar Eclipse 8.5.1b (https://www.solar-eclipse-genetics.org). Task-based function association analyses were based on NeuroSynth (Yarkoni et al., 2011) (https://neurosynth.org/). Full statistical scripts can be found at https://github.com/CNG-LAB/cngopen/tree/main/asymmetry_functional_gradients (copy archived at swh:1:rev:07d4a1a03267dac12ac8bfbccc8e09049cac9f31;path=/asymmetry_functional_gradients).

The following previously published data sets were used
    1. HCP S1200
    (2018) ConnectomeDB
    ID S1200. 1200 Subjects Data Release.

References

    1. Corballis MC
    (2009) The evolution and genetics of cerebral asymmetry
    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 364:867–879.
    https://doi.org/10.1098/rstb.2008.0232

Decision letter

  1. Saad Jbabdi
    Reviewing Editor; University of Oxford, United Kingdom
  2. Timothy E Behrens
    Senior Editor; University of Oxford, United Kingdom
  3. Saad Jbabdi
    Reviewer; University of Oxford, United Kingdom

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "Asymmetry of cortical functional hierarchy in humans and macaques suggests phylogenetic conservation and adaptation" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, including Saad Jbabdi as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Timothy Behrens as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

There are two important points to address in your revisions:

1) The theoretical point raised by reviewer 1 regarding interpretation of localised changes in the embedding space. This seems to be a fundamental limitation of the method.

2) The point raised by reviewer 2 about insight gained by this study.Reviewer #1 (Recommendations for the authors):

The paper is very well written, the question is interesting, and the analyses are innovative. However, I do have concerns about the overall approach. My main concern is about looking at asymmetries in the low dimensional representation of connectivity. A secondary concern has to do with looking at the parcellated connectome. I explain these concerns in succession below.

The first concern is to me quite a fundamental issue: looking at connectivity in a low dimensional space, that of the laplacian eigenvectors. There are two issues with this. The first one, which is less important than the second, is that the authors have a reference embedding to which they align other embeddings using a procrustes method with no scaling. While the 3D embedding is still optimally representing the connectivity (because distances don't change under rotations), we can no longer look at one axis at a time, which is what the authors do when they look at G1. In this case, G1 is representative of the connectivity of the reference matrix (LL), but not the others.

But even if the authors only projected their matrices onto a single G1 dimension with no procrustes (and only sign flipping if necessary), there is still a major issue. One implicit assumption of this whole approach is that if there is a change in connectivity somewhere in the original matrix, the same "nodes" of the matrix will change in the embedding. This is not the case. Any change in the original matrix, even if it is a single edge, will affect the positions of all the nodes in the embedding. That is because the embedding optimises a global loss function, not a local one.

To make this point clear, consider the following toy example. Say we have 4 brain regions A,B,C,D. Let us say that we have the following connectivity:

In the Left Hemisphere: A-B-C-D

In the Right Hemisphere: A-B=C-D

So the connection between B and C is twice as strong in the right hemi, and everything else remains the same.

The low dimensional embedding of both will look like this:

Left:

… A … B ……. C … D …

Right

A… … … B … C … … … D

Note how B,C are closer to each other in the RIGHT, but also that A,D have moved away from each other because the eigenvector has to have norm 1.

So if we were to calculate an asymmetry index, we would say that:

A is higher on the LEFT

B is higher on the RIGHT

C is higher on the LEFT

D is higher on the RIGHT

So we have found asymmetry in all of our regions. But in fact the only thing that has changed is the connection between B and C.

This illustrates the danger of using a global optimisation procedure (like low-dim embedding) to analyse and interpret local changes. One has to be very careful.

My second concern is about interpreting the brain asymmetry as differences in connectivity, as opposed to differences in other things like regional size. The authors use a parcellated approach, where presumably the parcels are left-right symmetric. If one area is actually larger in one hemisphere than in the other, the will manifest itself in the connectivity values. To mitigate this, it may be necessary to align the two hemispheres to each other (maybe using spherical registration) using connectivity prior to applying the parcellation.

Figure 1. Please explain what "explained variance" means. The gradients represent a low dimensional version of the connectivity matrix. they are not explaining variance?Reviewer #2 (Recommendations for the authors):

Using recently-developed functional gradient techniques, this study explored human brain hemispheric asymmetry. The functional gradient is a hot technique in recent years and has been applied to study brain asymmetries in two papers of 2021. Compared to previous studies, the current study further evaluated the degree of genetic control (heritability) and evolutionary conservation for such gradient asymmetries by using human twin data and monkey's fMRI data. These investigations are of value and do provide interesting data. However, it suffers from a lack of specific hypotheses/questions/motivations underlying all kinds of analyses, and the rich observational or correlational results seem not to offer significant improvement of theoretical understanding about brain asymmetries or functional gradient. In addition, given the limited number of twins in HCP project (for a heritability estimation), the limited number of monkeys (20 monkeys), and the relatively poor quality of monkeys' resting functional MRI data, the results and conclusion should be taken cautiously. Below are the concerns and suggestions.

The gradient from resting-state functional connectome has been frequently used but mainly at the group level. The current study essentially applied the gradient comparison (i.e., gradient score) at the individual level. Biological interpretation for individual gradient score at the parcel level as well as its comparability between individuals and between hemispheres should be resolved. This is the fundamental rationale underlying the whole analyses.

Only the first three gradients are used but why? What about the fourth gradient? Specific theoretical interpretation is needed. At the individual level, is it ensured that the first gradients of all individuals correspond to each other? In this study, it is unclear whether we should or should not care about the G2 and G3. The results of G2 and G3 showed up randomly to some degree.

The intra-hemispheric gradient is institutive. However, it is hard to understand what the inter-hemispheric gradient means. From the data perspective, yes you can do such gradient comparison between the LR and RL connectome but what does this mean? Why should we care about such asymmetry? From the introduction to the discussion, the authors simply showed the data of inter-hemispheric gradients without useful explanation. This issue should be solved.

When aligning intra-hemispheric gradient, choosing averaged LL mode as the reference may introduce systematic bias towards left hemisphere. Such an issue also applies to LR-RL gradient alignment as well as cross-species gradient alignment. This methodological issue should be solved.

The sample size of monkey (i.e., 20) is far less than human subjects (> 1000). Such limitation raises severe concern on the validity of the currently observed gradient asymmetry pattern in the monkey group, as well as the similarity results with human gradient asymmetry pattern. Despite the marginal significance of G1 inter-hemisphere gradient between humans and monkeys, I feel overall there is no convincingly meaningful similarity between these two species. However, the authors' discussion and conclusion are largely based on strong inter-species similarity in such asymmetry. The conclusion of evolutionary conservation for gradient asymmetry, therefore, is not well supported by the results.

For human gradient asymmetry, only t values were provided; For monkey gradient asymmetry, only Cohen-d values were provided. These two should be provided for both species.

Figure 3b, it is hard to believe that such a scatter plot can reach a significant correlation of R>0.3. In addition, such a scatter plot does not match the text (i.e., correlation between the "absolute" AI and heritability)

Figure S3, why should we care about these cross-gradient correlations?

More detailed description for fMRI post-processing for functional connectome and gradient analyses could be added in the supplementary information.

DK atlas is not a good validation parcellation for a functional MRI study like this.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Heritability and cross-species comparisons of asymmetry of human cortical functional organization" for further consideration by eLife. Your revised article has been evaluated by Timothy Behrens (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

Thanks for your replies to my comments, and sorry for the delay in getting this response to you.

Regarding my first comment, i.e. interpretation of a change in position along the gradients: I am not sure I understand your reply. You agreed that it is difficult to interpret these changes, given that they can represent changes occurring outside the region where the change is reported, but then the analysis you have done does not address this concern. Instead, you calculated some other measure (which I am not sure what it is as it is not well described) and reported the asymmetry index using this new measure. If this new measure is more interpretable, then why do you need to use gradients? What information from the gradients is useful for the study of asymmetries? And how can we interpret changes in positions along the gradients? Simply saying that "interpretation for asymmetry of areas is under a global context" seems to me like sweeping the issue under the rug.

Regarding the issue of using the Procrustes to the template and how that makes the gradients a worse representation of connectivity for the non-template matrices: I don't understand the reply here either. What is meant by joint alignment and how exactly does this address my concern?

If I may add a couple of additional points:

– I said in my original review that this was a well-written paper, but it looks like the writing has gotten worse in this revised paper. I am not sure why that is, but I really invite the authors to re-read the paper, particularly the new sections/paragraphs, and ensure that the arguments make sense (and I don't just mean the English).

– Some of the captions are way too short. They are often comprised of just a few words, which is ok for a caption "title", but not for a caption. A caption needs to explain what is shown avoiding reference to the main text.

– The code provided is poorly organised and not documented. I encourage the authors to improve on that.

https://doi.org/10.7554/eLife.77215.sa1

Author response

Reviewer #1 (Recommendations for the authors):

The paper is very well written, the question is interesting, and the analyses are innovative. However, I do have concerns about the overall approach. My main concern is about looking at asymmetries in the low dimensional representation of connectivity. A secondary concern has to do with looking at the parcellated connectome. I explain these concerns in succession below.

We thank the Reviewer for the appreciation of our work and the insightful comments, which we have addressed below.

The first concern is to me quite a fundamental issue: looking at connectivity in a low dimensional space, that of the laplacian eigenvectors. There are two issues with this. The first one, which is less important than the second, is that the authors have a reference embedding to which they align other embeddings using a procrustes method with no scaling. While the 3D embedding is still optimally representing the connectivity (because distances don't change under rotations), we can no longer look at one axis at a time, which is what the authors do when they look at G1. In this case, G1 is representative of the connectivity of the reference matrix (LL), but not the others.

But even if the authors only projected their matrices onto a single G1 dimension with no procrustes (and only sign flipping if necessary), there is still a major issue. One implicit assumption of this whole approach is that if there is a change in connectivity somewhere in the original matrix, the same "nodes" of the matrix will change in the embedding. This is not the case. Any change in the original matrix, even if it is a single edge, will affect the positions of all the nodes in the embedding. That is because the embedding optimises a global loss function, not a local one.

To make this point clear, consider the following toy example. Say we have 4 brain regions A,B,C,D. Let us say that we have the following connectivity:

In the Left Hemisphere: A-B-C-D

In the Right Hemisphere: A-B=C-D

So the connection between B and C is twice as strong in the right hemi, and everything else remains the same.

The low dimensional embedding of both will look like this:

Left:

… A … B ……. C … D …

Right

A… … … B … C … … … D

Note how B,C are closer to each other in the RIGHT, but also that A,D have moved away from each other because the eigenvector has to have norm 1.

So if we were to calculate an asymmetry index, we would say that:

A is higher on the LEFT

B is higher on the RIGHT

C is higher on the LEFT

D is higher on the RIGHT

So we have found asymmetry in all of our regions. But in fact the only thing that has changed is the connection between B and C.

This illustrates the danger of using a global optimisation procedure (like low-dim embedding) to analyse and interpret local changes. One has to be very careful.

We thank the Reviewer for the detailed description of the first concern. We agree that low-dimensional embeddings describe global embedding of local features, rather than local phenomena. Moreover, we indeed assume that the connectivity embedding of a given node gives us information about its position along ‘gradients’ relative to other nodes and their respective embedding. Thus, indeed, when a single node (node X) has a different connectivity profile in the right hemisphere relative to the left, this will also have some impact on the embeddings of all nodes showing a relevant (i.e., top 10%) connection to node X.

To evaluate whether asymmetry could be observed in average connectivity within functional networks, an alternative approach to measure asymmetry was taken by computing average connectivity within different functional networks. Following we compared the within-network connectivity between left and right. We have now added this conceptual analysis to our results robustness analysis section. In short, we observed that transmodal networks (DMN, FPN, and language network) showed higher connectivity in the left hemisphere but other networks showed higher connectivity in the right hemisphere. Thus, this indicates that observations made with respect to asymmetry of functional gradients are similar to those observed for within-network functional asymmetry between the left and right hemispheres. We have now detailed the outcome of this analysis in our Result section and Supplementary Materials.

Results, p.14.:

“As low-dimensional embedding is a global approach to summarize functional connectivity we reiterated our analysis by evaluating asymmetry of within network functional connectivity in the current sample. Observations made with respect to asymmetry of functional gradients are similar to those observed for within-network functional asymmetry between the left and right hemispheres.”

“To further explore functional connectivity asymmetry between left and right hemispheres, we calculated the LL within network FC and RR within network FC (Figure 2—figure supplement 5). It showed that connections in the left hemisphere and right hemisphere were relatively equal in the global scale. However, for the local differences, networks showed significant subtle leftward or rightward asymmetry (vis1: t = -5.203, P < 0.001; vis2: t = -22.593, P < 0.001; SMN: t = -8.262, P < 0.001; CON: t = -32.715, P < 0.001; DAN: t = -11.272, P < 0.001; Lan.: t = 33.827, P < 0.001; FPN: t = 24.439, P < 0.001; Aud.: t = 0.191, P = 0.849; DMN: t = 11.303, P < 0.001; PMN: t = -35.719, P < 0.001; VMN: t = -11.056, P < 0.001; OAN: t = 0.311, P = 0.756).”

Irrespectively, we have further highlighted that such a global interpretation for asymmetry of areas is still meaningful, given that a node is always placed in a global context. We have now further explained that our metrics give insights in local embedding of global phenomena in the introduction, p. 3.

Introduction, p. 3:

“These low-dimensional gradient embeddings describe global embedding of local features, rather than local phenomena. Thus, interpretation for asymmetry of areas is under a global context.”

My second concern is about interpreting the brain asymmetry as differences in connectivity, as opposed to differences in other things like regional size. The authors use a parcellated approach, where presumably the parcels are left-right symmetric. If one area is actually larger in one hemisphere than in the other, the will manifest itself in the connectivity values. To mitigate this, it may be necessary to align the two hemispheres to each other (maybe using spherical registration) using connectivity prior to applying the parcellation.

Thanks for this nice idea. We have now computed the differences of the mean rsfMRI connectome along the first gradient at the vertex level using 100 random subjects, as we have the data mapped to a symmetric template (fs_LR_32k), indicating that each vertex has a symmetric counterpart in the right hemisphere. Our results show left-right asymmetry as language/default mode-visual-frontoparietal vertices, which is consistent with the main results of the parcel-based approach. We have also added this response to the Supplementary materials.

Though overall findings are consistent, spherical registration may also have new issues. Total anatomical spatial symmetry may not provide functional comparability at the vertex level between left and right hemisphere. For example, during language tasks in the current sample, the activated frontal region in the left hemisphere is larger than the activated contralateral region in the right hemisphere. In the current study, we aimed to evaluate asymmetry between functionally and structurally homologous regions, as described by the Glasser atlas. In case of the resting state fMRI data, we used the region-wise symmetric multimodal parcellation (Glasser et al., 2016). This parcellation ensures the functional contralateral regions in both hemispheres. A previous study (Williams et al., 2021) investigated the structural and functional asymmetry in newborn infants. They used spherical registration (make fs_LR symmetric) for structural asymmetry but not for functional asymmetry. As such spheric registration may hide functional information, we think spherical registration may be more suitable for structural studies.

To address the concern regarding the alignment of hemispheres, we used joint alignment for LL and RR to compare the results between this and the Procrustes alignment technique (Pearson r=0.930, P_spin<0.001), Figure2—figure supplement 7 is the figure of asymmetry along the principal gradient (upper: joint alignment, below: Procrustes alignment) indicating convergence between both approaches. We have reported this information in the Supplementary Materials.

Lastly, we do agree that parcel size might be an important issue influencing the asymmetry pattern. To test for such an effect, we performed the correlation between the rank of parcel size (left-right)/(left+right) and rank of asymmetry index. It suggests only a small insignificant correlation along G1 (Spearman r_intra=0.130, P_spin=0.105; Spearman r_inter=0.130, P_spin=0.084). Of note, there is a systematic difference in parcel size as a function of sensory-association hierarchy, indicating that the link between parcel-size and asymmetry may vary as a function of sensory vs associative regions.

Figure 1. Please explain what "explained variance" means. The gradients represent a low dimensional version of the connectivity matrix. they are not explaining variance?

We thank the Reviewer for this question. The “explained variance” along the y-axis is the contribution of each eigenvector to the whole-brain connectivity variance decomposed but not the eigen-variance along (G1, G2, and G3 that are presented in the 3D scatter of Figure 1). We have revised the legend of Figure 1d about the “Variance %”.

Figure 1d, legend: “(d) Gradient template using the group-level gradient of LL. Dots represent parcels and were colored according to Cole-Anticevic networks. The decomposition scatter in the right below side depicts x-axis (number of eigenvectors) and y-axis (the contribution of each eigenvector to the total)”

Reviewer #2 (Recommendations for the authors):

Using recently-developed functional gradient techniques, this study explored human brain hemispheric asymmetry. The functional gradient is a hot technique in recent years and has been applied to study brain asymmetries in two papers of 2021. Compared to previous studies, the current study further evaluated the degree of genetic control (heritability) and evolutionary conservation for such gradient asymmetries by using human twin data and monkey's fMRI data. These investigations are of value and do provide interesting data. However, it suffers from a lack of specific hypotheses/questions/motivations underlying all kinds of analyses, and the rich observational or correlational results seem not to offer significant improvement of theoretical understanding about brain asymmetries or functional gradient. In addition, given the limited number of twins in HCP project (for a heritability estimation), the limited number of monkeys (20 monkeys), and the relatively poor quality of monkeys' resting functional MRI data, the results and conclusion should be taken cautiously. Below are the concerns and suggestions.

We thank the Reviewer for the evaluation of our work and the helpful suggestions.

The gradient from resting-state functional connectome has been frequently used but mainly at the group level. The current study essentially applied the gradient comparison (i.e., gradient score) at the individual level. Biological interpretation for individual gradient score at the parcel level as well as its comparability between individuals and between hemispheres should be resolved. This is the fundamental rationale underlying the whole analyses.

We thank the Reviewer for this remark, and are happy to provide further rationale for using and comparing individual gradients scores to evaluate individual variation in asymmetry and associated heritability. Though gradients from resting-state functional connectivity have been frequently used at the group level, various studies have also studied individual differences. For example, using linear mixed models to compare gradient scores between left and right across subjects (Liang et al., 2021), applying the individual gradient scores to compare disease and controls (Dong et al., 2020, 2021; Hong et al., 2019; Park et al., 2021), and link individual hippocampal gradients to memory recollection (Przeździk et al., 2019). Together, these studies show individual variations of local gradients, indicating changes in node centrality and hubness (Hong et al., 2019), and connectivity profile distance (Y. Wang et al., 2021). Of note, low-dimensional embeddings describe global embedding of local features, rather than local phenomena. Thus, interpretation for asymmetry of areas is under a global context. The biological interpretation for individual gradients would be to what degree the system segregated and integrated has changed patterns of ongoing neural activity (Mckeown et al., 2020). It reflects that individuals have different functional boundaries between anatomical regions. Whereas individual neurons are embedded under the global-local boundaries through a cortical wiring space consisting of intricate long- and short-range white matter fibers (Paquola et al., 2020).

Introduction, p. 4:

“We applied the individual gradient scores to study the asymmetry, consistent with prior studies (Gonzalez Alam et al., 2021; Liang et al., 2021). Individual variation along the gradients reflects a global change across subjects in the functional connectome integration and segregation, and it is under genetic control (Valk et al., 2021). Moreover, to what degree the system segregated and integrated relates to patterns of ongoing neural activity (Mckeown et al., 2020), and different individuals have different functional boundaries between anatomical regions.”

Results, p. 5:

“Next, individual gradients were computed for each subject and the four different FC modes and aligned to the template gradients with Procrustes rotation. It rotates a matrix to maximum similarity with a target matrix minimizing sum of squared differences. As noted, Procrustes matching was applied without a scaling factor so that the reference template only matters for matching the order and direction of the gradients. Therefore, it allows comparison between individuals and hemispheres. The individual mean gradients showed high correlation with the group gradients LL (all Pearson r > 0.97, P spin < 0.001).”

Only the first three gradients are used but why? What about the fourth gradient? Specific theoretical interpretation is needed. At the individual level, is it ensured that the first gradients of all individuals correspond to each other? In this study, it is unclear whether we should or should not care about the G2 and G3. The results of G2 and G3 showed up randomly to some degree.

In the current study we focused on the principal gradient in the main analysis, given its association with sensory-transmodal hierarchy, microstructure, and evolutionary alterations (Margulies et al., 2016; Paquola et al., 2019; Xu et al., 2020).

Conversely, gradient 2 reflects the dissociation between visual and sensory-motor networks and gradient 3 is linked to task-positive, control, versus ‘default’ and sensory-motor regions. We analyzed asymmetry and its heritability of the first three gradients (explaining respectively 23.3%, 18.1%, and 15.0% of the variance of the rsFC matrix). However, we extracted the first ten gradients to maximize the degree of fit (Margulies et al., 2016; Mckeown et al., 2020). We have now also shown G4-10 mean asymmetry results as a supplementary figure. To ensure correspondence of gradients across individuals, we aligned the individual gradients to the group level template with Procrustes rotation. Procrustes rotation rotates a matrix to maximum similarity with a target matrix minimizing sum of squared differences. The approach is typically used in comparison of ordination results and is particularly useful in comparing alternative solutions in multidimensional scaling. Figure S1 shows the mean gradients across subjects of each FC mode, which is close to the Figure 1D template gradient space.

Results, p. 5:

“The current study analyzed asymmetry and its heritability of the first three gradients explaining most variance (Figure 1d). As they all have reasonably well described functional associations (G1: unimodal-transmodal gradient with 24.1%, G2: somatosensory-visual gradient with 18.4%, G3: multi-demand gradient with 15.1%). However, given we extracted ten gradients to maximize the degree of fit 26,52. We stated mean asymmetry of G4-10 in Figure 1—figure supplement 1.”

The intra-hemispheric gradient is institutive. However, it is hard to understand what the inter-hemispheric gradient means. From the data perspective, yes you can do such gradient comparison between the LR and RL connectome but what does this mean? Why should we care about such asymmetry? From the introduction to the discussion, the authors simply showed the data of inter-hemispheric gradients without useful explanation. This issue should be solved.

We are happy to further clarify. The LR and RL connectivity reflects cross-hemispheric functional signal interaction via corpus callosum, whose structural asymmetry is usually studied (Karolis et al., 2019). Such intra-hemispheric connections, compared to the inter-hemispheric connections, have been suggested to reflect the inhibition of corpus callosum, and underlie hemispheric specialization. Different information relies on hemispheric specialization (e.g., visual, motor, and crude information) and/or inter-hemispheric information transfer (e.g., language, reasoning, and attention) (Gazzaniga, 2000). To clarify and motivate the analysis of both intra- and inter-hemispheric asymmetry in functional gradients, we have now added further detail in the introduction, p. 5.

Here is text:

Introduction, p. 4.

“The full FC matrix contains both intra-hemispheric and inter-hemispheric connections. Intra-hemispheric connections, compared to the inter-hemispheric connections, have been suggested to reflect the inhibition of corpus callosum and may underlie hemispheric specializations involving language, reasoning, and attention. Conversely, inter-hemispheric connectivity may reflect information transfer between hemispheres, for example a wide range of modal and motor information, and crude information concerning spatial locations 48. Previous studies have reported intra-hemispheric FC to study gradient asymmetry 6,38. By having the callosum related to association white matter fibers, one hemisphere could develop for new functions while the other hemisphere could continue to perform the previous functions for both hemispheres 48. Therefore, in addition to the intra-hemispheric FC gradients, we depicted the inter-hemispheric FC, which is abnormal in patients with schizophrenia 23,49 and autism 24.”

as well as Discussion, p. 16

“Conversely, the transmodal frontoparietal network was located at the apex of rightward preference, possibly suggesting a right-ward lateralization of cortical regions associated with attention and control and ‘default’ internal cognition 62,63. The observed dissociation between language and control networks is also in line with previous work suggesting an inverse pattern of language and attention between hemispheres 3,64. Such patterns may be linked to inhibition of corpus callosum 65, promoting hemispheric specialization. It has been suggested that such inter-hemispheric connections set the stage for intra-hemispheric patterns related to association fibers 48. Future research may relate functional asymmetry directly to asymmetry in underlying structure to uncover how different white-matter tracts contribute to asymmetry of functional organization.”

and Discussion, p.18

“Though overall intra- and inter-hemispheric connectivity showed a strong spatial overlap in humans, we also observed marked differences between both metrics across our analysis. For example, although we found both intra- and inter-hemispheric differences in gradient organization to be heritable, only for intra-hemispheric asymmetry we found a correspondence between degree of asymmetry and degree of heritability. Similarly comparing asymmetry observed in human data to functional gradient asymmetry in macaques, we only observed spatial patterning of asymmetry was conserved for intra-hemispheric connections. Whereas intra-hemispheric asymmetry relates to association fibers, commissural fibers underlie inter-hemispheric connections 77 It has been suggested that there is a trade-off within and across mammals of inter- and intra-hemispheric connectivity patterns to conserve the balance between grey and white-matter 76. Consequently, differences in asymmetry of both ipsi- and contralateral functional connections may be reflective of adjustments in this balance within and across species. Secondly, previous research studying intra- and inter-hemispheric connectivity and associated asymmetry has indicated a developmental trajectory from inter- to intra-hemispheric organization of brain functional connectivity, varying from unimodal to transmodal areas 78,79. It is thus possible that a reduced correspondence of asymmetry and heritability in humans, as well as lack of spatial similarities between humans and macaques for inter-hemispheric connectivity may be due to the age of both samples (young adults in humans, adolescents in macaques). Further research may study inter- and intra-hemispheric asymmetry in functional organization as a function of development in both species to further disentangle heritability and cross-species conservation and adaptation.”

When aligning intra-hemispheric gradient, choosing averaged LL mode as the reference may introduce systematic bias towards left hemisphere. Such an issue also applies to LR-RL gradient alignment as well as cross-species gradient alignment. This methodological issue should be solved.

We thank the Reviewer for raising this point. Indeed, we also used RR as reference, the results were virtually identical. We have stated this in the Results, p. 13. Regarding the cross-species alignment, we averaged the left and right hemispheres to reduce the systematic bias. It showed that the correlation and comparison results remained robust. Now we have updated the method and corresponding results (p.10). Here is the text:

Results (p.15):

“We also set the RR FC gradients as reference, the first three of which explained 22.8%, 18.8%, and 15.9% of total variance. We aligned each individual to this reference. It suggested all results were virtually identical (Pearson r > 0.9, P spin < 0.001).”

Results (p.10):

“To reduce a possible systematic hemispheric bias during the cross-species alignment, we averaged the left and right hemisphere. We found that the macaque and macaque-aligned human AI maps of G1 were correlated positively for intra-hemispheric patterns (Pearson r = 0.345, P spin = 0.030). For inter-hemispheric patterns, we didn’t observe a significant association (Pearson r = -0.029, P spin = 0.858)”

The sample size of monkey (i.e., 20) is far less than human subjects (> 1000). Such limitation raises severe concern on the validity of the currently observed gradient asymmetry pattern in the monkey group, as well as the similarity results with human gradient asymmetry pattern. Despite the marginal significance of G1 inter-hemisphere gradient between humans and monkeys, I feel overall there is no convincingly meaningful similarity between these two species. However, the authors' discussion and conclusion are largely based on strong inter-species similarity in such asymmetry. The conclusion of evolutionary conservation for gradient asymmetry, therefore, is not well supported by the results.

We agree with your comments. Although it is a small sample compared to humans, in NHP studies, it is a relatively decent sample size (most of the studies have N<10). Of note, recent work suggested that the individual variation pattern can be captured using 4 subjects in both human and macaques (Ren et al., 2021).

To overcome potential overinterpretation of our findings, we have now changed the title to a more descriptive format:

“Heritability and cross-species comparisons of asymmetry of human cortical functional organization”

And further detailed findings already in the Abstract;

“These asymmetries were heritable in humans and, for intra-hemispheric asymmetry of functional connectivity, showed similar spatial distributions in humans and macaques, suggesting phylogenetic conservation.”

We have pointed out the small sample size in the limitation. Please find the text below:

Discussion, p. 18:

“Due to the small sample size of macaques, it is important to be careful when interpreting our observations regarding asymmetry in macaques, and its relation to asymmetry patterning observed in humans. Therefore, further study is needed to evaluate the asymmetry patterns in macaques using large datasets 53,79

And nuanced the conclusion, p.19:

“This asymmetry was heritable and, in the case of organization of intra-hemispheric connectivity, showed spatial correspondence between humans and macaques. At the same time, functional asymmetry was more pronounced in language networks in humans relative to macaques, suggesting adaptation.”

For human gradient asymmetry, only t values were provided; For monkey gradient asymmetry, only Cohen-d values were provided. These two should be provided for both species.

Thanks for the comment. For the humans, we have now provided the Cohen’s d map as a supplementary figure.

Figure 3b, it is hard to believe that such a scatter plot can reach a significant correlation of R>0.3. In addition, such a scatter plot does not match the text (i.e., correlation between the "absolute" AI and heritability)

Thanks for pointing this out. In the previous version of the figure, we wanted to show whether a region showing relatively high heritability has a left or right-ward asymmetry, and computed their relationship using absolute values of asymmetry. We have now revised the corresponding figure, and added further detail in the results paragraph, please see below and Results Figure 3 section, p.9.

Results Figure 3, p. 9:

“To assess whether regions showing higher asymmetry had an increased heritability of G1, we plotted our cortical maps of asymmetry along those reporting heritability (Figure 3b). For the correlation between the absolute asymmetry index and heritability (Figure 3b small scatter), gradients of the intra-hemispheric FC patterns were significant (Pearson r = 0.245, P spin = 0.005) but gradients of the inter-hemispheric FC were not (Pearson r = 0.055, P spin = 0.613).”

Figure S3, why should we care about these cross-gradient correlations?

Thanks for your comment and we apologize when the motivation was not clear, originally we intended to transparently display whether there was any relationship between asymmetry G1-3 in humans and macaques. However, we agree with the Reviewer that this analysis is unclear, and moves away from the main question of the work. Thus, we have now removed this cross-gradient correlations figure and focused in particular on the G1 cross-species correlation.

More detailed description for fMRI post-processing for functional connectome and gradient analyses could be added in the supplementary information.

We are happy to further detail fMRI post-processing for functional connectome and gradient analysis, please find below. in the Methods;

FC, p.21:

“All rs-fMRI data underwent HCP’s minimal preprocessing 80 and were coregistered using a multimodal surface matching algorithm (MSMAll) 83 to the HCP template 32k_LR surface space. The template consists of 32,492 total vertices per hemisphere (59,412 excluding the medial wall). Cortical time series were averaged within a previously established multi-modal parcellation schemes: for humans the 360-parcel Glasser atlas (180 per hemisphere) 54 and the 182-parcel Markov atlas (91 per hemisphere) for macaques 56. To compute the functional connectivity (FC), time-series of cortical parcels were correlated pairwise using the Pearson product moment and then Fisher’s z-transformed in human and macaque data, separately. Individual FC maps were also averaged across four different rs-fMRI sessions for humans ([LR1], [LR2], [RL1], and [RL2]). We computed the FC in four different patterns, both for human and macaque data: FC within the left and right hemispheres (LL intra-hemisphere, RR intra-hemisphere), from the left to right hemisphere (LR inter-hemisphere) and from the right to left hemisphere (RL, inter-hemisphere).”

Connectivity gradients, p.21:

“Next we employed the nonlinear dimensionality reduction technique 26 to generate the group level gradients of the mean LL FC across individuals. We then set the group-level gradients as the template and aligned each individual gradient with Procrustes rotation to the template. Finally, the comparative individual functional gradients of each FC pattern were assessed. All steps were accomplished in the Python package Brainspace 27. In brief, the algorithm estimates a low-dimensional embedding from a high-dimensional affinity matrix. Along these low-dimensional axes, or gradients, cortical nodes that are strongly interconnected, by either many suprathreshold edges or few very strong edges, are closer together. Nodes with little connectivity similarly are farther apart. Regions having similar connectivity profiles are embedded together along the gradient axis. The name of this approach, which belongs to the family of graph Laplacians, is derived from the equivalence of the Euclidean distance between points in the diffusion embedded mapping 25–27. It is controlled by a single parameter α, which controls the influence of the density of sampling points on the manifold (α = 0, maximal influence; α = 1, no influence). On the basis of the previous work 26, we followed recommendations and set α = 0.5, a choice that retains the global relations between data points in the embedded space and has been suggested to be relatively robust to noise in the covariance matrix.”

“The input of the analysis was the FC matrix, which was cut off at 90% similar to previous studies 26. The current study selected the first three FC LL gradients (G1, G2, and G3) that explained 24.1%, 18.4%, and 15.1% of total variance in humans, as well as 18.9%, 15.2%, and 12.8% of total variance in macaques”

DK atlas is not a good validation parcellation for a functional MRI study like this.

We agree with the Reviewer. However, given that various structural MRI studies on brain asymmetry (e.g., Kong et al., 2018, 2022; Sha et al., 2021) have reported results using DK atlas, we have put this figure on the Figure 2 supplements for comparison of results and completeness with respect to this literature.

References

Dong, D., Luo, C., Guell, X., Wang, Y., He, H., Duan, M., Eickhoff, S. B., and Yao, D. (2020). Compression of Cerebellar Functional Gradients in Schizophrenia. Schizophrenia Bulletin, 46(5), 1282–1295. https://doi.org/10.1093/schbul/sbaa016

Dong, D., Yao, D., Wang, Y., Hong, S.-J., Genon, S., Xin, F., Jung, K., He, H., Chang, X., Duan, M., Bernhardt, B. C., Margulies, D. S., Sepulcre, J., Eickhoff, S. B., and Luo, C. (2021). Compressed sensorimotor-to-transmodal hierarchical organization in schizophrenia. Psychological Medicine, 1–14. https://doi.org/10.1017/S0033291721002129

Gazzaniga, M. S. (2000). Cerebral specialization and interhemispheric communication: Does the corpus callosum enable the human condition? Brain, 123(7), 1293–1326. https://doi.org/10.1093/brain/123.7.1293

Glasser, M. F., Coalson, T. S., Robinson, E. C., Hacker, C. D., Harwell, J., Yacoub, E., Ugurbil, K., Andersson, J., Beckmann, C. F., Jenkinson, M., Smith, S. M., and Van Essen, D. C. (2016). A multi-modal parcellation of human cerebral cortex. Nature, 536(7615), 171–178. https://doi.org/10.1038/nature18933

Hong, S.-J., Vos de Wael, R., Bethlehem, R. A. I., Lariviere, S., Paquola, C., Valk, S. L., Milham, M. P., Di Martino, A., Margulies, D. S., Smallwood, J., and Bernhardt, B. C. (2019). Atypical functional connectome hierarchy in autism. Nature Communications, 10(1), 1022. https://doi.org/10.1038/s41467-019-08944-1

Karolis, V. R., Corbetta, M., and Thiebaut de Schotten, M. (2019). The architecture of functional lateralisation and its relationship to callosal connectivity in the human brain. Nature Communications, 10(1), 1417. https://doi.org/10.1038/s41467-019-09344-1

Kong, X.-Z. et al. Mapping cortical brain asymmetry in 17,141 healthy individuals worldwide via the ENIGMA Consortium. PNAS 115, E5154–E5163 (2018).

Kong, X.-Z. et al. Mapping brain asymmetry in health and disease through the ENIGMA consortium. Hum Brain Mapp 43, 167–181 (2022).

Liang, X., Zhao, C., Jin, X., Jiang, Y., Yang, L., Chen, Y., and Gong, G. (2021). Sex-related human brain asymmetry in hemispheric functional gradients. NeuroImage, 229, 117761. https://doi.org/10.1016/j.neuroimage.2021.117761

Margulies, D. S., Ghosh, S. S., Goulas, A., Falkiewicz, M., Huntenburg, J. M., Langs, G., Bezgin, G., Eickhoff, S. B., Castellanos, F. X., Petrides, M., Jefferies, E., and Smallwood, J. (2016). Situating the default-mode network along a principal gradient of macroscale cortical organization. Proceedings of the National Academy of Sciences, 113(44), 12574–12579. https://doi.org/10.1073/pnas.1608282113

Mckeown, B., Strawson, W. H., Wang, H.-T., Karapanagiotidis, T., Vos de Wael, R., Benkarim, O., Turnbull, A., Margulies, D., Jefferies, E., McCall, C., Bernhardt, B., and Smallwood, J. (2020). The relationship between individual variation in macroscale functional gradients and distinct aspects of ongoing thought. NeuroImage, 220, 117072. https://doi.org/10.1016/j.neuroimage.2020.117072

Paquola, C., Wael, R. V. D., Wagstyl, K., Bethlehem, R. A. I., Hong, S.-J., Seidlitz, J., Bullmore, E. T., Evans, A. C., Misic, B., Margulies, D. S., Smallwood, J., and Bernhardt, B. C. (2019). Microstructural and functional gradients are increasingly dissociated in transmodal cortices. PLOS Biology, 17(5), e3000284. https://doi.org/10.1371/journal.pbio.3000284

Park, B., Hong, S.-J., Valk, S. L., Paquola, C., Benkarim, O., Bethlehem, R. A. I., Di Martino, A., Milham, M. P., Gozzi, A., Yeo, B. T. T., Smallwood, J., and Bernhardt, B. C. (2021). Differences in subcortico-cortical interactions identified from connectome and microcircuit models in autism. Nature Communications, 12(1), 2225. https://doi.org/10.1038/s41467-021-21732-0

Przeździk, I., Faber, M., Fernández, G., Beckmann, C. F., and Haak, K. V. (2019). The functional organisation of the hippocampus along its long axis is gradual and predicts recollection. Cortex, 119, 324–335. https://doi.org/10.1016/j.cortex.2019.04.015

Ren, J., Xu, T., Wang, D., Li, M., Lin, Y., Schoeppe, F., Ramirez, J. S. B., Han, Y., Luan, G., Li, L., Liu, H., and Ahveninen, J. (2021). Individual Variability in Functional Organization of the Human and Monkey Auditory Cortex. Cerebral Cortex, 31(5), 2450–2465. https://doi.org/10.1093/cercor/bhaa366

Sha, Z. et al. The genetic architecture of structural left–right asymmetry of the human brain. Nat Hum Behav 5, 1226–1239 (2021).

Wang, Y., Royer, J., Park, B., Wael, R. V. de, Larivière, S., Tavakol, S., Rodriguez-Cruces, R., Paquola, C., Hong, S.-J., Margulies, D. S., Smallwood, J., Valk, S. L., Evans, A. C., and Bernhardt, B. C. (2021). Long-range connections mirror and link microarchitectural and cognitive hierarchies in the human brain (p. 2021.10.25.465692). https://www.biorxiv.org/content/10.1101/2021.10.25.465692v1

Williams, L. Z. J., Fitzgibbon, S. P., Bozek, J., Winkler, A. M., Dimitrova, R., Poppe, T., Schuh, A., Makropoulos, A., Cupitt, J., O’Muircheartaigh, J., Duff, E. P., Cordero-Grande, L., Price, A. N., Hajnal, J. V., Rueckert, D., Smith, S. M., Edwards, A. D., and Robinson, E. C. (2021). Structural and functional asymmetry of the neonatal cerebral cortex (p. 2021.10.13.464206). bioRxiv. https://doi.org/10.1101/2021.10.13.464206

Xu, T., Nenning, K.-H., Schwartz, E., Hong, S.-J., Vogelstein, J. T., Goulas, A., Fair, D. A., Schroeder, C. E., Margulies, D. S., Smallwood, J., Milham, M. P., and Langs, G. (2020). Cross-species functional alignment reveals evolutionary hierarchy within the connectome. NeuroImage, 223, 117346. https://doi.org/10.1016/j.neuroimage.2020.117346

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #1 (Recommendations for the authors):

Thanks for your replies to my comments, and sorry for the delay in getting this response to you.

Regarding my first comment, i.e. interpretation of a change in position along the gradients: I am not sure I understand your reply.

We thank the Reviewer for these important concerns. Please find our revised response to the original question below.

Q1.1_Rv1: The first concern is to me quite a fundamental issue: looking at connectivity in a low dimensional space, that of the laplacian eigenvectors. There are two issues with this. The first one, which is less important than the second, is that the authors have a reference embedding to which they align other embeddings using a procrustes method with no scaling. While the 3D embedding is still optimally representing the connectivity (because distances don't change under rotations), we can no longer look at one axis at a time, which is what the authors do when they look at G1. In this case, G1 is representative of the connectivity of the reference matrix (LL), but not the others. But even if the authors only projected their matrices onto a single G1 dimension with no procrustes ( and only sign flipping if necessary) […]

We now project the matrices in a single G1 dimension without Procrustes alignment, and with flipped signs where needed, as suggested by the Reviewer. Comparing the gradient asymmetry with Procrustes alignment to the gradient without alignment resulted in virtually identical results for the HCP sample ( r intra-hemisphere = 0.956 , r inter-hemisphere = 0.843). At the same time, comparing unaligned and aligned gradients in the UK Biobank sample, we find that the alignment improves the similarity to the pattern observed in HCP (aligned r intra-hemisphere = 0.592 , non-aligned r intra-hemisphere = 0.487, aligned r inter-hemisphere = 0.384, non-aligned r inter-hemisphere = 0.162). We agree with the Reviewer that the Procrustes alignment procedure may create somewhat of a mixture between two vectors. However, we consider the impact of this operation minor and that in general Procrustus alignment comes with important benefits, namely to be able to make comparisons of multiple vectors across participants and samples. For example, using Procrustes alignment has been shown to have beneficial effects on the reproducibility and stability of gradients across techniques, parameters and individuals (Hong et al., 2020).

We now include an explicit comparison of Procrustes and non-aligned gradients in our supplemental materials, p. 15.

“To evaluate potential downstream effects of alignment to our results, we compared the gradient asymmetry with Procrustes alignment to the gradient without alignment. This resulted in virtually identical results for the HCP sample ( r intra-hemisphere = 0.956 , r inter-hemisphere = 0.843, Figure 2—figure supplement 7 ). At the same time, comparing unaligned and aligned gradients in the UKB sample, we found that the alignment improved the similarity to the pattern observed in HCP (aligned r intra-hemisphere = 0.592, non-aligned r intra-hemisphere = 0.487, aligned r inter-hemisphere = 0.384, non-aligned r inter-hemisphere = 0.162, Figure 2-figure supplement 8).”

is higher on the LEFT

is higher on the RIGHT

is higher on the LEFT

is higher on the RIGHT

So we have found asymmetry in all of our regions. But in fact the only thing that has changed is the connection between B and C.

This illustrates the danger of using a global optimisation procedure (like low-dim embedding) to analyse and interpret local changes. One has to be very careful.

We thank the Reviewer for this extensive and thorough comment. To overcome potential differences in normalization of eigenvectors that may occur when computing LL_RR and LR_RL gradients separately we computed the gradient of LL_RR and LR_RL in the same model, under the assumption that if the homologous regions in the left and right hemisphere would have the same connectivity, they would have the same gradient loading. However, if their connectivity pattern is different, they would have a different loading yet differences would be normalized equally for LL_RR and LR_RL as they are part of the same model.

We find that, along the principal gradient, the observed normalized asymmetric map is highly similar to the non-normalized map used in the main analyses for intra-hemispheric (Pearson r = 0.956). Conversely, for the inter-hemispheric (Pearson r = 0.531) asymmetry patterns are still consistent, yet we find the similarity reduced. It is possible the difference between intra- and inter-hemispheric correspondence relates to more global differences in strength of connectivity comparing LR to RL FC, as reported also in (Raemaekers et al., 2018) resulting in more widespread differences between inter-hemispheric patterns of both embedding procedures.

We have now included this analysis in the supplementary results, p. 15 and as a supplementary Figure.

“Moreover, to overcome potential normalization biases associated with creating one gradient for each hemisphere, we performed an alternative analysis to create a gradient of the left and right hemisphere together. This assumes that regions with similar connectivity profiles have comparable loading in the gradient framework. Indeed, along the principal gradient, the observed normalized asymmetric map was highly similar to the non-normalized map used in the main analyses for the intra-hemispheric (Pearson r = 0.956) and inter-hemispheric (Pearson r = 0.531) asymmetry patterns (Figure 2-figure supplement 9). It is possible the difference between intra- and inter-hemispheric correspondence relates to more global differences in strength of connectivity comparing LR to RL FC, as reported also in the article 7 resulting in more widespread differences between inter-hemispheric patterns of both embedding procedures.”

To further display the underlying differences in connectivity profiles we furthermore projected the top 10% of connectivity patterns which were the basis for dimensionality reduction techniques used in the main analyses on the cortex. Thus, any change that may alter potential gradient loadings would fall within the top 10% of a regions’ connectivity profiles. We selected the asymmetric parcel (No.25: Peri-Sylvian language area) of and displayed their left-right difference in top 10% connectivity profiles between left and right hemisphere. We have now added this example to the supplementary figures, and refer to it in the results, p.8.

"The differences in gradient loadings (parcel No.25: Peri-Sylvian language area) reflect differences in connectivity profiles (top 10%) between LL versus RR, or LR versus RL respectively (Figure 2—figure supplement 1)."

You agreed that it is difficult to interpret these changes, given that they can represent changes occurring outside the region where the change is reported, but then the analysis you have done does not address this concern.

We thank the Reviewer for this comment, and hope that both our answers to previous questions (Q1.Rv1 and Q1.2_Rv1) have created further clarity. Please find further answers and clarifications below.

Instead, you calculated some other measure (which I am not sure what it is as it is not well described) and reported the asymmetry index using this new measure. If this new measure is more interpretable, then why do you need to use gradients? What information from the gradients is useful for the study of asymmetries? And how can we interpret changes in positions along the gradients? Simply saying that "interpretation for asymmetry of areas is under a global context" seems to me like sweeping the issue under the rug.

We thank the Reviewer for raising these important points. Indeed, we have now revisited this question and provided updated analyses which provide a more fitting answer to the raised concerns. Overall we believe that using a gradient framework to study asymmetry of functional brain organization has various benefits and provides us with novel insights into the asymmetric organization of functional connectivity that can be integrated within the wider literature of cortical organization. First of all, gradients provide a synoptic framework to capture smooth variations of connectivity patterns across the cortical mantle. Such differences have been linked to graph-theoretical markers such as degree centrality (Hong et al., 2019) , and dynamics (Park et al., 2021) as well as connectivity distance (Hong et al., 2019; Wang et al., 2022). Moreover, the principal gradient provides a coordinate framework spanning the cortex and reflecting the geodesic distance between primary and default regions, and relates to cortical microstructure and associated processing hierarchies (Huntenburg et al., 2018). In doing so, and in contrast to clustering or network-based approaches, the gradient framework provides a spatial ordering of functional brain networks, placing them along a gradual axis of connectivity variation reaching from sensory to transmodal areas. In the context of asymmetry of gradient loadings this would mean that a given region with a significant left-ward asymmetry along the first gradient (sensory-to-transmodal) has a connectivity profile more similar to the transmodal anchor than in the left hemisphere relative to the right. Consequently these regions are placed at different positions along the cortical hierarchy, providing novel insights in the system-level variations of the asymmetric brain. This data-driven framework enabled us to evaluate heritability, i.e. to what extent genetic factors impact variation across individuals in asymmetric organization of the cortex, and compare patterns between humans and macaques. Moreover, our framework can, in future studies, be used to evaluate overlaps between asymmetry of local cortical structure with the functional (and structural) asymmetric embedding of a given region.

To better understand the asymmetry between hemispheres, previous work has also employed the gradients framework (Gonzalez Alam et al., 2022; Liang et al., 2021) and found inter-hemispheric functional connectivity profile differences in frontoparietal and default mode networks. In the current work, we further optimized the approach by considering individual variation and using an atlas that considers contralateral homologous regions and is based on the current sample (Glasser, 2016). Indeed, contrary to a network level account of asymmetry, the gradient framework helps us to describe the asymmetry of a large-scale functional coordinate system spanning sensory to transmodal anchors using a data-driven approach. We have now further highlighted these arguments in the Introduction, p.3.

“Gradients provide a synoptic framework to capture smooth variations of connectivity patterns across the cortical mantle. They describe variations in genetic patterning 33,3435 , functional processes 26,32,36 , and are observed across species 34,37,38. Gradients have been linked to graph-theoretical markers such as degree centrality 39 and microcircuit dynamics 40 as well as connec-tivity distance 39,41. Moreover, the principal gradient describes the geodesic distance between primary and default regions, and relates to cortical microstructure and associated processing hierarchies 42. In doing so, and in contrast to clustering or network-based approaches, the gradi-ent framework provides a spatial ordering of functional brain networks, placing them along a gradual axis of connectivity variation reaching from sensory to transmodal areas. In the context of asymmetry of gradient loadings this would mean that a given region with a significant left-ward asymmetry along the first gradient (sensory-to-transmodal) has a connectivity profile more similar to the transmodal anchor in the left hemisphere relative to the right. Consequently, these regions are placed at different positions along the cortical hierarchy, providing novel in-sights concerning the system-level variations in the asymmetric brain.”

Regarding the issue of using the Procrustes to the template and how that makes the gradients a worse representation of connectivity for the non-template matrices: I don't understand the reply here either. What is meant by joint alignment and how exactly does this address my concern?

We apologize for the confusion. Joint alignment is implemented based on spectral embedding. Here the embedding, rather than using the affinity matrices individually, is based on the joint affinity matrix (Vos de Wael et al., 2020). However, as we now have updated our reply to the original question to provide a more fitting answer we believe this analysis is no longer relevant and have again removed its outcome from the supplementary materials.

Reference

Gonzalez Alam, T. R. del J., Mckeown, B. L. A., Gao, Z., Bernhardt, B., Vos de Wael, R., Margulies, D. S., Smallwood, J., and Jefferies, E. (2022). A tale of two gradients: Differences between the left and right hemispheres predict semantic cognition. Brain Structure and Function , 227 (2), 631–654. https://doi.org/10.1007/s00429-021-02374-w

Hong, S.-J., Vos de Wael, R., Bethlehem, R. A. I., Lariviere, S., Paquola, C., Valk, S. L., Milham, M. P., Di Martino, A., Margulies, D. S., Smallwood, J., and Bernhardt, B. C. (2019). Atypical functional connectome hierarchy in autism. Nature Communications, 10 (1), 1022. https://doi.org/10.1038/s41467-019-08944-1

Hong, S.-J., Xu, T., Nikolaidis, A., Smallwood, J., Margulies, D. S., Bernhardt, B., Vogelstein, J., and Milham, M. P. (2020). Toward a connectivity gradient-based framework for reproducible biomarker discovery. NeuroImage, 223 , 117322. https://doi.org/10.1016/j.neuroimage.2020.117322

Huntenburg, J. M., Bazin, P.-L., and Margulies, D. S. (2018). Large-Scale Gradients in Human Cortical Organization. Trends in Cognitive Sciences, 22 (1), 21–31. https://doi.org/10.1016/j.tics.2017.11.002

Liang, X., Zhao, C., Jin, X., Jiang, Y., Yang, L., Chen, Y., and Gong, G. (2021). Sex-related human brain asymmetry in hemispheric functional gradients. NeuroImage, 229, 117761. https://doi.org/10.1016/j.neuroimage.2021.117761

Park, B., Hong, S.-J., Valk, S. L., Paquola, C., Benkarim, O., Bethlehem, R. A. I., Di Martino, A., Milham, M. P., Gozzi, A., Yeo, B. T. T., Smallwood, J., and Bernhardt, B. C. (2021). Differences in subcortico-cortical interactions identified from connectome and microcircuit models in autism. Nature Communications, 12 (1), 2225. https://doi.org/10.1038/s41467-021-21732-0

Raemaekers, M., Schellekens, W., Petridou, N., and Ramsey, N. F. (2018). Knowing left from right: Asymmetric functional connectivity during resting state. Brain Structure and Function, 223 (4), 1909–1922. https://doi.org/10.1007/s00429-017-1604-y

Vos de Wael, R., Benkarim, O., Paquola, C., Lariviere, S., Royer, J., Tavakol, S., Xu, T., Hong, S.-J., Langs, G., Valk, S., Misic, B., Milham, M., Margulies, D., Smallwood, J., and Bernhardt, B. C. (2020). BrainSpace: A toolbox for the analysis of macroscale gradients in neuroimaging and connectomics datasets. Communications Biology, 3 (1), 1–10. https://doi.org/10.1038/s42003-020-0794-7

Wang, Y., Royer, J., Park, B., Vos de Wael, R., Larivière, S., Tavakol, S., Rodriguez-Cruces, R., Paquola, C., Hong, S.-J., Margulies, D. S., Smallwood, J., Valk, S. L., Evans, A. C., and Bernhardt, B. C. (2022). Long-range functional connections mirror and link microarchitectural and cognitive hierarchies in the human brain. Cerebral Cortex, bhac172. https://doi.org/10.1093/cercor/bhac172

https://doi.org/10.7554/eLife.77215.sa2

Article and author information

Author details

  1. Bin Wan

    1. Otto Hahn Group Cognitive Neurogenetics, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
    2. International Max Planck Research School on Neuroscience of Communication: Function, Structure, and Plasticity (IMPRS NeuroCom), Leipzig, Germany
    3. Department of Cognitive Neurology, University Hospital Leipzig and Faculty of Medicine, University of Leipzig, Leipzig, Germany
    4. Institute of Neuroscience and Medicine (INM-7: Brain and Behavior), Research Centre Jülich, Jülich, Germany
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Validation, Visualization, Methodology, Writing - original draft, Project administration, Writing – review and editing
    For correspondence
    wanb.psych@outlook.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9077-3354
  2. Şeyma Bayrak

    1. Otto Hahn Group Cognitive Neurogenetics, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
    2. Department of Cognitive Neurology, University Hospital Leipzig and Faculty of Medicine, University of Leipzig, Leipzig, Germany
    3. Institute of Neuroscience and Medicine (INM-7: Brain and Behavior), Research Centre Jülich, Jülich, Germany
    Contribution
    Validation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  3. Ting Xu

    Center for the Developing Brain, Child Mind Institute, New York, United States
    Contribution
    Data curation, Writing – review and editing
    Competing interests
    No competing interests declared
  4. H Lina Schaare

    1. Otto Hahn Group Cognitive Neurogenetics, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
    2. Institute of Neuroscience and Medicine (INM-7: Brain and Behavior), Research Centre Jülich, Jülich, Germany
    Contribution
    Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4259-0793
  5. Richard AI Bethlehem

    Department of Psychiatry, University of Cambridge, Cambridge, United Kingdom
    Contribution
    Data curation, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0714-0685
  6. Boris C Bernhardt

    McConnell Brain Imaging Centre, Montréal Neurological Institute and Hospital, McGill University, Montréal, Canada
    Contribution
    Funding acquisition, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9256-6041
  7. Sofie L Valk

    1. Otto Hahn Group Cognitive Neurogenetics, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
    2. Institute of Neuroscience and Medicine (INM-7: Brain and Behavior), Research Centre Jülich, Jülich, Germany
    3. Institute of Systems Neuroscience, Heinrich Heine University Düsseldorf, Düsseldorf, Germany
    Contribution
    Conceptualization, Data curation, Supervision, Funding acquisition, Validation, Methodology, Project administration, Writing – review and editing
    For correspondence
    valk@cbs.mpg.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2998-6849

Funding

Max-Planck-Gesellschaft

  • Sofie L Valk

Sick Kids Foundation (NI17-039)

  • Boris C Bernhardt

Natural Sciences and Engineering Research Council of Canada (Discovery-1304413)

  • Boris C Bernhardt

Canadian Institutes of Health Research (FDN154298)

  • Boris C Bernhardt

Azrieli Center for Autism Research

  • Boris C Bernhardt

Canada First Research Excellence Fund

  • Boris C Bernhardt
  • Sofie L Valk

International Max Planck Research School on Neuroscience of Communication: Function, Structure, and Plasticity

  • Bin Wan

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to thank the various contributors to the open access databases that our data was downloaded from. Funding: HCP data were provided by the Human Connectome Project, Washington University, the University of Minnesota, and Oxford University Consortium (Principal Investigators: David Van Essen and Kamil Ugurbil;1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University. Additional personnel support provided by the Center for the Developing Brain at the Child Mind Institute, as well as NIMH R01MH081218, R01MH083246, and R21MH084126. Project support is also provided by the NKI Center for Advanced Brain Imaging (CABI), the Brain Research Foundation (Chicago, IL), and the Stavros Niarchos Foundation. This study was supported by the Deutsche Forschungsgemeinschaft (DFG, EI 816/21–1), the National Institute of Mental Health (R01-MH074457), the Helmholtz Portfolio Theme "Supercomputing and Modeling for the Human Brain'' and the European Union’s Horizon 2020 Research and Innovation Program under Grant Agreement No. 785,907 (HBP SGA2). SLV was supported by Max Planck Gesellschaft (Otto Hahn award). BCB acknowledges support from the SickKids Foundation (NI17-039), the National Sciences and Engineering Research Council of Canada (NSERC; Discovery-1304413), CIHR (FDN154298), Azrieli Center for Autism Research (ACAR), an MNI-Cambridge collaboration grant, and the Canada Research Chairs program. Last, this work was funded in part by Helmholtz Association’s Initiative and Networking Fund under the Helmholtz International Lab grant agreement InterLabs-0015, and the Canada First Research Excellence Fund (CFREF Competition 2, 2015–2016) awarded to the Healthy Brains, Healthy Lives initiative at McGill University, through the Helmholtz International BigBrain Analytics and Learning Laboratory (HIBALL), including SLV and BCB. BW was supported by the International Max Planck Research School on Neuroscience of Communication: Function, Structure, and Plasticity (IMPRS NeuroCom).

Ethics

The current research complies with all relevant ethical regulations as set by The Independent Research Ethics Committee at the Medical Faculty of the Heinrich-Heine-University of Duesseldorf (study number 2018-317).

Senior Editor

  1. Timothy E Behrens, University of Oxford, United Kingdom

Reviewing Editor

  1. Saad Jbabdi, University of Oxford, United Kingdom

Reviewer

  1. Saad Jbabdi, University of Oxford, United Kingdom

Publication history

  1. Preprint posted: November 4, 2021 (view preprint)
  2. Received: January 20, 2022
  3. Accepted: July 28, 2022
  4. Accepted Manuscript published: July 29, 2022 (version 1)
  5. Accepted Manuscript updated: August 11, 2022 (version 2)
  6. Version of Record published: August 16, 2022 (version 3)

Copyright

© 2022, Wan et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,683
    Page views
  • 630
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Bin Wan
  2. Şeyma Bayrak
  3. Ting Xu
  4. H Lina Schaare
  5. Richard AI Bethlehem
  6. Boris C Bernhardt
  7. Sofie L Valk
(2022)
Heritability and cross-species comparisons of human cortical functional organization asymmetry
eLife 11:e77215.
https://doi.org/10.7554/eLife.77215

Further reading

    1. Neuroscience
    Frédéric Roux, George Parish ... Simon Hanslmayr
    Research Article Updated

    Theta and gamma oscillations in the medial temporal lobe are suggested to play a critical role for human memory formation via establishing synchrony in neural assemblies. Arguably, such synchrony facilitates efficient information transfer between neurons and enhances synaptic plasticity, both of which benefit episodic memory formation. However, to date little evidence exists from humans that would provide direct evidence for such a specific role of theta and gamma oscillations for episodic memory formation. Here, we investigate how oscillations shape the temporal structure of neural firing during memory formation in the medial temporal lobe. We measured neural firing and local field potentials in human epilepsy patients via micro-wire electrode recordings to analyze whether brain oscillations are related to co-incidences of firing between neurons during successful and unsuccessful encoding of episodic memories. The results show that phase-coupling of neurons to faster theta and gamma oscillations correlates with co-firing at short latencies (~20–30 ms) and occurs during successful memory formation. Phase-coupling at slower oscillations in these same frequency bands, in contrast, correlates with longer co-firing latencies and occurs during memory failure. Thus, our findings suggest that neural oscillations play a role for the synchronization of neural firing in the medial temporal lobe during the encoding of episodic memories.

    1. Neuroscience
    Sarah M Lurie, James E Kragel ... Joel L Voss
    Research Article Updated

    Hippocampal-dependent memory is thought to be supported by distinct connectivity states, with strong input to the hippocampus benefitting encoding and weak input benefitting retrieval. Previous research in rodents suggests that the hippocampal theta oscillation orchestrates the transition between these states, with opposite phase angles predicting minimal versus maximal input. We investigated whether this phase dependence exists in humans using network-targeted intracranial stimulation. Intracranial local field potentials were recorded from individuals with epilepsy undergoing medically necessary stereotactic electroencephalographic recording. In each subject, biphasic bipolar direct electrical stimulation was delivered to lateral temporal sites with demonstrated connectivity to hippocampus. Lateral temporal stimulation evoked ipsilateral hippocampal potentials with distinct early and late components. Using evoked component amplitude to measure functional connectivity, we assessed whether the phase of hippocampal theta predicted relatively high versus low connectivity. We observed an increase in the continuous phase–amplitude relationship selective to the early and late components of the response evoked by lateral temporal stimulation. The maximal difference in these evoked component amplitudes occurred across 180 degrees of separation in the hippocampal theta rhythm; that is, the greatest difference in component amplitude was observed when stimulation was delivered at theta peak versus trough. The pattern of theta-phase dependence observed for hippocampus was not identified for control locations. These findings demonstrate that hippocampal receptivity to input varies with theta phase, suggesting that theta phase reflects connectivity states of human hippocampal networks. These findings confirm a putative mechanism by which neural oscillations modulate human hippocampal function.