1. Neuroscience
Download icon

Macro-connectomics and microstructure predict dynamic plasticity patterns in the non-human primate brain

  1. Sean Froudist-Walsh
  2. Philip GF Browning
  3. James J Young
  4. Kathy L Murphy
  5. Rogier B Mars
  6. Lazar Fleysher
  7. Paula L Croxson  Is a corresponding author
  1. Icahn School of Medicine at Mount Sinai, United States
  2. Friedman Brain Institute, Icahn School of Medicine at Mount Sinai, United States
  3. National Institute of Mental Health, United States
  4. Newcastle University, United Kingdom
  5. John Radcliffe Hospital, University of Oxford, United Kingdom
  6. Radboud University, The Netherlands
Research Article
Cite this article as: eLife 2018;7:e34354 doi: 10.7554/eLife.34354
8 figures, 4 tables, 1 data set and 1 additional file

Figures

Bilateral hippocampal lesions.

(A) T2-weighted hypersignal 6 days after surgery indicating local inflammation in the hippocampus; overlap is shown for the five monkeys. (B) Sketch of hippocampal size based on histology (Nissl stained sections) overlaid on atlas sections. The unlesioned hippocampal volume is shown in red. Overlap of remaining hippocampal volume is shown for the five monkeys indicating shrinkage of the hippocampus bilaterally in all monkeys. (C-D) Results of deformation-based morphometry analysis showing atrophy of the hipppocampus (C) 3 months after the lesion and (D) 12 months after the lesion.

https://doi.org/10.7554/eLife.34354.003
Anatomical and functional predictors of plasticity.

(A-B) Neuron and non-neuronal cell densities were mapped from Collins et al., 2010. (C) Hubness was calculated as each area’s projection onto the first principal component of node strength and network participation coefficient data. (D) Pre-lesion hippocampal functional connectivity was strongly correlated with anatomical connectivity derived from the CoCoMac tract-tracing atlas (r = 0.54, p = 2.2×10−7) and enhanced versions of the CoCoMac Atlas (r = 0.60, p = 3.6×10−9). The hippocampus was most strongly connected to ventral temporal lobe structures..

https://doi.org/10.7554/eLife.34354.004
Changes in network participation are strongly predicted by pre-lesion anatomy and functional connectivity.

(A) Most brain regions showed a drop in network participation over the acute stage. (B-C) The degree to which individual brain regions reduced their network participation over the acute stage was well predicted by their pre-lesion connectivity to the hippocampus and the extent to which they acted as hubs in the pre-lesion network (‘hubness’). (B) A scatter plot of the acute stage changes to the network participation coefficient for each brain regions, compared to model predictions. Brain regions are coloured according to their pre-lesion connectivity with the hippocampus (compare with Figure 2D). (D-F) As in (A-C), but for chronic stage changes to network participation. Note that areas with a higher non-neuronal cell density showed the greatest increase in the network participation coefficient over the chronic stage. Note that (D) shows the overall within network participation coefficient changes for the chronic stage, while the model predictions and data shown in (E-F) corresponded to the residual chronic stage changes to within the network participation coefficient, after regressing out acute stage changes. * signifies that these predictors were significant and included in the final model.

https://doi.org/10.7554/eLife.34354.005
Pre-lesion hippocampal connectivity is associated with a rise in within-module connectivity over the chronic stage.

(A) The pattern of acute stage increases and decreases to within-module connectivity. (B-C) The degree to which individual brain regions changed their within-module connectivity over the acute stage was significantly associated with the extent to which they acted as hubs in the pre-lesion network (‘hubness’). Scatter plot in (B) shows brain regions coloured according to their pre-lesion connectivity with the hippocampus (compare with Figure 2D). (D-F) As in (A-C), but for chronic stage changes to within-module connectivity. Note that (D) shows the overall within module-connectivity changes for the chronic stage, while the model predictions and data shown in (E-F) corresponded to the residual chronic stage changes to within module-connectivity, after regressing out acute stage changes. * signifies that these predictors were significant and included in the final model.

https://doi.org/10.7554/eLife.34354.006
Grey matter loss at the chronic stage was most prominent in cortical regions that were strongly connected to the hippocampus.

(A) The pattern of acute stage increases and decreases to cortical grey matter volume. (B-C) No significant predictors of acute stage cortical grey matter changes were identified. (D) Whole brain voxelwise analysis revealed very limited volumetric decreases in the medial septum, amygdala and dorsal premotor cortex. (E) As in (A), but for chronic stage changes to cortical grey-matter volume. (E-F). The degree to which individual cortical regions changed their grey-matter volume over the chronic stage was significantly associated with the extent to which they were functionally connected to the hippocampus before the lesion. Scatter plot in (B) shows brain regions coloured according to their pre-lesion connectivity with the hippocampus (compare with Figure 2D). Note that (E) shows the overall within module-connectivity changes for the chronic stage, while the model predictions and data shown in (F-G) corresponded to the residual chronic stage changes to within module-connectivity, after regressing out acute stage changes. * signifies that this predictor was significant and included in the final model. (H) Whole brain voxelwise analysis revealed a larger range of decreases in grey-matter volume over the chronic stage. There were also volumetric increases in the cerebellum, midbrain and premotor cortex. These results did not survive multiple comparisons correction.

https://doi.org/10.7554/eLife.34354.007
Effects of hippocampal lesions on module structure.

The most consistent pre-operatively defined modules are shown anatomically (A) and using a force-directed graph representation (B) where highly functionally connected brain regions are plotted close together. Before the lesion, the parieto-occipital module (orange) is highly connected to the three other modules. At 3 months post-lesion, the network looks largely similar, although there may have been some dispersion. At 12 months post-lesion however, the parieto-occipital module is completely dispersed, with some dispersion of other modules. (C-D) We quantified this dispersion by the mean change in within-module functional connectivity for each module. There is a significant effect of module on node dispersion over both the acute (C – pre-lesion vs. 3 months post-lesion) and chronic (D – 3 months vs 12 months post-lesion) stages.

https://doi.org/10.7554/eLife.34354.008
Author response image 1
Connectivity strength distribution at different stages.
Author response image 2
Showing the acute and chronic changes in overall node strength, and its relationship with the four independent variables.

Tables

Table 1
Remaining volume of tissue in each lesioned monkey (calculated relative to atlas volumes from Nissl-stained histological sections registered to atlas sections) and lesion extent expressed as a percentage (1-(remaining volume/normal volume)).
https://doi.org/10.7554/eLife.34354.010
MonkeyLeft hemisphere remaining volumeRight hemisphere remaining volumeBilateral remaining volumeLeft hemisphere lesion %Right hemisphere lesion %Total lesion %
Atlas268144268665536809
Mean14937912189527127444.2954.6349.46
E18449814604733054531.1945.6438.42
M1297418487721461851.6268.4160.02
N15953515878731832240.5040.9040.70
S947698951218428164.6666.6865.67
T17835413025430860833.4951.5242.51
Table 2
Details of monkeys and surgeries.
https://doi.org/10.7554/eLife.34354.009
T1-weightedResting-state
MonkeyGroupPre3 month1 yearPre3 month1 year
ELesionXXXX
MLesionXXXXXX
NLesionXXXX
SLesionXXXXX
TLesionXXXXXX
CControlXX
LControlXX
WControlXX
Total854644
Table 3
Lesion volumes calculated from T2-weighted hypersignal relative to whole hippocampal volume for each monkey. All monkeys received two lesion attempts except monkey E.
https://doi.org/10.7554/eLife.34354.011
HippocampusLeftRight
MonkeyLesion attemptsVolumeLesion%VolumeLesion%VolumeLesion%
E1821.38244.1329.72429.2594.8822.10392.13149.2538.06
M21019.75563.3855.25516.63190.5036.87503.13372.8874.11
N21161.38179.5015.46607.63109.3818.00553.7570.1312.66
S2979.24706.4972.15484.12398.3782.29495.12308.1262.23
T2937.38690.6373.68442.88364.0082.19494.50326.6366.05
Table 4
List of regional map abbreviations and corresponding brain areas.
https://doi.org/10.7554/eLife.34354.012
Regional map abbreviationBrain area
A1Primary auditory cortex
A2Secondary auditory cortex
IaAnterior insula
IpPosterior insula
AmygAmygdala
CCaAnterior cingulate cortex
CCpPosterior cingulate cortex
CCrRetrosplenial cortex
CCsSubgenual cingulate cortex
FEFFrontal eye field
GGustatory area
HCHippocampus
M1Primary motor cortex
PFCclCentrolateral prefrontal cortex
PFCdlDorsolateral prefrontal cortex
PFCdmDorsomedial prefrontal cortex
PFCmMedial prefrontal cortex
PFCoiIntermediate orbital prefrontal cortex
PFColOrbitolateral prefrontal cortex
PFComOrbitomedial prefrontal cortex
PFCvlVentrolateral prefrontal cortex
PFCpolPolar prefrontal cortex
PHCParahippocampal cortex
PMCdlDorsolateral premotor cortex
PMCmMedial (supplementary) premotor cortex
PMCvlVentrolateral premotor cortex
S1Primary somatosensory cortex
S2Secondary somatosensory cortex
PCiInferior parietal cortex
PCipCortex of the intraparietal sulcus
PCmMedial parietal cortex
PCsSuperior parietal cortex
TCcCentral temporal cortex
TCiInferior temporal cortex
TCsSuperior temporal cortex
TCpolPolar temporal cortex
TCvVentral temporal cortex
V1Primary visual cortex
V2Secondary visual cortex
VACdDorsal anterior visual cortex
VACvVentral anterior visual cortex

Data availability

Data are available to download from the INDI PRIMatE Data Exchange (Milham et al., 2018): https://www.nitrc.org/account/login.php?return_to=http://fcon_1000.projects.nitrc.org/indi/PRIMEdownloads.html as the Mount Sinai Philips Achieva 3T dataset. Users will first be prompted to log on to NITRC and will need to register with the 1000 Functional Connectomes Project website on NITRC (http://fcon_1000.projects.nitrc.org/indi/PRIME/mssm1.html) to gain access to the PRIME-DE datasets. The code used for analysis has been made available on GitHub: https://github.com/seanfw/froudist-walsh-et-al-elife-2018 (copy archived at https://github.com/elifesciences-publications/froudist-walsh-et-al-elife-2018).

The following data sets were generated
  1. 1
    INDI PRIMatE Data Exchange
    1. D Margulies
    2. M Milham
    3. C Schroeder
    (2017)
    Mount Sinai Philips Achieva 3T.

Additional files

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)

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

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