Research ArticleNEUROSCIENCE

Spatiotemporal ontogeny of brain wiring

See allHide authors and affiliations

Science Advances  12 Jun 2019:
Vol. 5, no. 6, eaav9694
DOI: 10.1126/sciadv.aav9694
  • Fig. 1 Brain connectomes are characterized by interregional connectivity strength heterogeneity and exhibit characteristic global network topology.

    Heterogeneity of interregional connectivity strength is parsimoniously described by the wiring cost and connectional homophily principles (right). Schematic depiction illustrates characteristic global network topology features observed in diverse brain connectomes (left). For simplicity, directionality of connections is omitted.

  • Fig. 2 Homophily, physical distance, and strength heterogeneity of empirical brain connectomes.

    Increased homophily entails increased connectivity strength. Increased physical distance entails decreased connectivity strength. Depicted R2 values are derived from univariate regression. See the “Homophily and wiring cost principles” section for results and R2 values concerning multivariate regression. Note that the correlation, and thus the reported R2 values, between the strength of connections, physical distance, and homophily was computed on the nonbinned data and that the binning was performed for aiding visualization. Drosophila brain drawing from (6).

  • Fig. 3 Developmental modeling approach.

    (A) Neurodevelopmental gradients were simulated in a synthetic 2D brain. Each surface unit was characterized by a time window that indicates at a given time point the probability of a neuronal population migrating to each surface unit. Each time window is shaped by the distance of each surface unit from the root(s), that is, origin(s), of the neurodevelopmental gradients. For instance, the surface unit close to the neurodevelopmental origin (petrol green) is more probable to be populated earlier than the surface unit further from the neurodevelopmental origin (magenta). (B) Heterochronous and spatially ordered ontogeny of synthetic connectomes. (C) Heterochronous, spatially random, and (D) tautochronous ontogeny of synthetic connectomes. (E) Creation of the synthetic connectome. Note that time units depicted in (A) and (B) denote “developmental ticks” occurring in a time frame from the onset of the simulation of development and connectivity formation until the end of the simulation (for details, see the “Modeling neurodevelopmental gradients and connectivity formation” section).

  • Fig. 4 Interregional wiring principles captured by the synthetic connectomes.

    Summary of R2 values for the predictions of empirical connectivity from parameters calculated exclusively from the synthetic connectomes. Values for R2 are depicted for the (A) Drosophila, (B) mouse, (C) macaque monkey, and (D) human connectomes when considering the homophily or wiring cost principles separately or simultaneously. Note the better fit of the heterochronously generated connectomes when considering simultaneously the wiring cost and homophily principles. Note that the predictions from model parameters estimated exclusively from the synthetic connectomes result in R2 values very close to R2 values obtained when using exclusively the empirical data (gray bars), thus replicating a high percentage of the empirically observed relations between connectivity strength, homophily, and distance across vertebrate and invertebrate brains. Drosophila brain drawing from (6).

  • Fig. 5 Network topology morphospace of synthetic connectomes and classification of empirical connectomes.

    (A) Heterochronously and tautochronously generated connectomes are separated on the basis of their global network properties across the first component. Core size (max clique size) and characteristic path length (average shortest path length) are the major network metrics differentiating the groups. (B) Posterior probabilities indicate that the global network topology signature of all empirical connectomes is more reminiscent of connectomes with a heterochronous ontogeny. Drosophila brain drawing from (6).

  • Fig. 6 Global network topology metrics for the synthetic and empirical connectomes.

    The percentages below each bar denote the percentage of the values obtained for each network metric in relation to the empirical values. Thus, values over or below 100% denote overestimation or underestimation of the network metric, respectively. Drosophila brain drawing from (6).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/6/eaav9694/DC1

    Fig. S1. Neurodevelopmental gradients in the developing empirical and synthetic brain.

    Fig. S2. AIC values for the different simulation scenarios.

    Fig. S3. AIC values for comparison of the models.

    Fig. S4. Examples of actual and predicted strength of connections for the different models across brain connectomes.

    Fig. S5. Examples of the projections of the empirical brain connectomes to the morphospace defined by the global network topology of the synthetic brain connectomes.

    Fig. S6. AIC values for the similarity of the global network topology of empirical and synthetic connectomes.

    Table S1. F-test values for the predictions of the strength of connections of the empirical connectomes from parameters estimated exclusively from the synthetic connectomes.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Neurodevelopmental gradients in the developing empirical and synthetic brain.
    • Fig. S2. AIC values for the different simulation scenarios.
    • Fig. S3. AIC values for comparison of the models.
    • Fig. S4. Examples of actual and predicted strength of connections for the different models across brain connectomes.
    • Fig. S5. Examples of the projections of the empirical brain connectomes to the morphospace defined by the global network topology of the synthetic brain connectomes.
    • Fig. S6. AIC values for the similarity of the global network topology of empirical and synthetic connectomes.
    • Table S1. F-test values for the predictions of the strength of connections of the empirical connectomes from parameters estimated exclusively from the synthetic connectomes.

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article