Research ArticleNEUROSCIENCE

Intrinsic coupling modes reveal the functional architecture of cortico-tectal networks

See allHide authors and affiliations

Science Advances  07 Aug 2015:
Vol. 1, no. 7, e1500229
DOI: 10.1126/sciadv.1500229
  • Fig. 1 Experimental setup.

    (A) Schematic diagram of the custom-designed μECoG array. Sixty-four electrodes of 250-μm diameter were distributed across three separate polyimide fingers and arranged in a hexagonal grid (1.5-mm inter-electrode spacing). Holes were cut into the polyimide foils in the space between electrodes to allow for the placement of linear silicon probes. (B) A photo from the surgical implantation of the μECoG array. The general area for SC penetrations is shown by a blue box. Black lines indicate the lateral sulcus (LAT) and the suprasylvian sulcus (SSY). (C) Schematic diagram illustrating the placement of dual-shank 32-channel silicon probes in the SC. Probes were placed such that neural data could be acquired from both superficial (blue) and deep (green) layers of the SC simultaneously. (D) Schematic illustration of the placement of linear silicon probes in the visual cortex. Single-shank 32-channel probes (100-μm inter-electrode spacing) were advanced into the cortex through small holes in the μECoG array. Probes were advanced until the most superficial contacts were just above the pial surface, such that we recorded, in a single penetration, from superficial and deep visual cortex simultaneously. SZ, stratum zonale; SGS, stratum griseum superficiale; SO, stratum opticum; SGI, stratum griseum intermediale; PAG, periaqueductal gray.

  • Fig. 2 Dynamics and large-scale topography of cortico-cortical and cortico-tectal amplitude envelope correlation.

    (A) Top: Population-averaged amplitude envelope correlation computed between μECoG contact pairs separated by varying distances. Note that correlations are widely distributed in slow (~0.7 Hz), delta (~3 Hz), and spindle (8 to 15 Hz) frequency ranges and that amplitude envelopes are minimally correlated for frequencies above 120 Hz. Middle: Population-averaged amplitude envelope correlation computed between intracortical and μECoG recording sites. Bottom: Population-averaged cortico-tectal amplitude envelope correlation computed for SC recording contacts located in superficial (blue) and deep (red) SC layers. Note the peaks in cortico-tectal amplitude correlation for delta, spindle, and gamma frequencies. (B) Average cortical topography of cortico-tectal amplitude correlation for delta, spindle, and gamma (30 to 45 Hz) carrier frequencies. Maps are plotted for both superficial and deep SC layers. To compare functional coupling to anatomy, we plotted the density of tectally projecting neurons across the cortical surface. Anatomical data were adapted with permission from Fig. 1 in (13). (C) Spatial correlation of anatomical connectivity (B, bottom) and functional coupling topographies across all frequencies. The z score of correlation coefficients was estimated by computing the spatial correlation of randomly scrambled amplitude correlation topographies and anatomical data. Analyses for superficial and deep layers are plotted in blue and red, respectively.

  • Fig. 3 High-frequency LFP amplitude envelopes are correlated between cortex and SC.

    (A) Simultaneously recorded SC, intracortical, and μECoG signals from one example recording session. No relationship between signals is immediately visible upon inspection of LFPs (top). However, high-frequency amplitude envelopes display burst-like fluctuations that appeared to occur synchronously. (B) Cortical topography of high-frequency amplitude envelope correlation for the seed SC electrode displayed in (A). Note the region-specific amplitude correlation for lateral visual and suprasylvian cortical areas. (C) Cortical topography of high-frequency amplitude envelope correlation for the intracortical recording site displayed in (A). Note the presence of correlated high-frequency activity in cortical regions immediately surrounding the position of the intracortical recording site (marked as *). (D) Strength of high-frequency amplitude correlation between all SC and intracortical channel pairs. Note the presence of a cluster of correlated channels in the center of the SC probe and the lower third of the intracortical probe.

  • Fig. 4 Large-scale topography of high-frequency cortico-tectal amplitude envelope correlation.

    (A) Left: Average cortical topography of LFP amplitude correlation between superficial SC and μECoG array. Note the strongest amplitude correlation over posterior visual cortex. Middle: Average cortical topography of deep SC to μECoG LFP amplitude correlation. Note the presence of strong amplitude correlation in visual and suprasylvian cortical areas. Amplitude correlation effects extend anterior-medially along the suprasylvian gyrus toward a separate cluster of strong correlation in suprasylvian cortex. Right: Plot of the density of tectally projecting neurons across the cortical surface. Data were adapted with permission from (13). (B) Population-averaged (±SEM) strength of superficial SC-μECoG (left) and deep SC-μECoG (middle) high-frequency amplitude envelope correlation for different cortical areas. A map of the areal parcellation used in this analysis is shown on the right. Note the presence of strong correlation in early visual cortical areas for superficial SC-μECoG channel pairs. In contrast, deep SC-μECoG amplitude correlation was more widespread, encompassing visual, suprasylvian, and posterior parietal areas. (C) Left: Incidence of significant SC-μECoG high-frequency amplitude correlation as a function of SC depth. Right: Population-averaged spike/noise ratio in response to visual flash stimulation (flash onset at 0 s). Note that significant amplitude correlation with cortex was most prominent in intermediate/deep SC layers.

  • Fig. 5 High-frequency amplitude envelope correlation reflects synchronized cortico-tectal spiking activity.

    (A) SC STA spectrograms for two different μECoG contacts from an example recording session. The location of the contacts is shown in the panel on the left. Both spectrograms were computed using spiking data from the same SC seed contact. The strength of cortico-tectal amplitude correlation for the SC seed electrode is displayed as a heat map. Note the presence of an increase of μECoG signal power for frequencies above 20 Hz time-locked to SC spiking activity for the channel pair that showed strong power correlation (right, bottom panel). In contrast, the weakly correlated channel pair displayed no spike-triggered change in signal power (right, top panel). (B) Population mean (±SD) STA spectrograms. SC-μECoG channel pairs displaying significant high-frequency amplitude correlation are plotted in blue, whereas all others are plotted in red. (C) Correlation of amplitude envelope coupling and STA power for all frequency-frequency combinations. Note that high-frequency coupling is correlated to the strength of STA signals in frequencies above ~3 Hz. (D) Example spike cross-correlation histograms for SC-intracortical channel pairs that displayed strong (blue) and weak (red) high-frequency amplitude envelope correlation, respectively. Note the large peak around ~0 ms for the correlated channel pair and the lack of structure for the uncorrelated channel pair. (E) Scatter plot of the strength of amplitude correlation plotted against the probability of synchronous spiking for all SC-intracortical channel pairs. The correlation coefficient and related P value are shown as an inset. (F) Correlation of amplitude envelope coupling and synchronized spiking activity as a function of frequency (±SD). Note that coupling in high LFP frequencies specifically correlates with synchronized cortico-tectal spiking activity.

  • Fig. 6 Comparison of spontaneous and stimulus-induced high-frequency neural dynamics.

    (A) Drifting grating induced power changes for representative SC (left) and μECoG (middle) recording sites. The gray bar above each spectrogram indicates the duration of drifting grating stimulation. Note the broadband increase in high-frequency power for both SC and μECoG recording sites. Right: A scatter plot of the strength of spontaneous SC-μECoG versus drifting grating-induced SC-μECoG high-frequency amplitude correlation. Note that most data points lie below the 1:1 line, indicating that SC-μECoG channel pairs display stronger amplitude correlation during spontaneous activity. (B) Left: μECoG and SC visual spatial receptive fields from one example recording session. Note that this SC-μECoG channel pair displayed considerable visual spatial receptive field overlap as well as strong high-frequency amplitude envelope correlation during spontaneous activity. Right: Histogram displaying the distribution of visual receptive field correlation for SC-μECoG channel pairs that displayed significant high-frequency amplitude correlation (blue) and insignificant correlation (red). *P < 0.01.

  • Fig. 7 The dynamics of spontaneous SC neural activity are dominated by the phase of slow and spindle oscillations.

    (A) Population-averaged (±SEM) SC spike PLVs calculated using the phase of cortical (left) and local SC oscillations (right). The dashed line in each plot illustrates the level of significance (P < 0.01). For SC-μECoG channel pairs, significantly correlated and uncorrelated channel pairs are plotted separately. The black bar at the bottom of the SC-μECoG plot indicates the area in which correlated and uncorrelated curves are significantly different (P < 0.01). Note that spiking in amplitude-correlated channel pairs is more strongly locked to the phase of slow cortical oscillations. In addition, SC spiking activity is locked to the phase of local oscillations at the spindle frequency (~10 Hz). (B) Population-averaged cross-frequency phase-amplitude spectrograms calculated using the phase of μECoG oscillations and the amplitude of SC signals. Significantly correlated SC-μECoG channel pairs are shown on the left, uncorrelated channel pairs in the middle, and the difference between correlated and uncorrelated on the right. Note that SC-μECoG channel pairs that display strong high-frequency amplitude correlation also show significantly stronger coupling of SC oscillations above 8 Hz to the phase of slow cortical oscillations.

  • Fig. 8 The dependency of SC spiking activity to the phase of cortical slow oscillations.

    (A) Example of μECoG signal filtered at the slow oscillation frequency with a raster plot containing spiking data from several SC channels. Note that the bursting behavior of SC neurons appears to be phase-locked to the trough of the filtered slow wave. (B) Spike-phase histograms for the two channels marked (blue and green) in (A). Note that neurons from the more superficial recording contact prefer a slightly earlier phase than the deeper recording contact. (C) Population-averaged spike-phase histogram as a function of SC depth. Histograms were compiled using SC spiking activity and the phase of μECoG slow oscillations, and centered on the slow oscillatory phase corresponding to cortical up states, where a relative phase of 0 indicates the center of the up state located close to the trough of cortical slow oscillations. The strength with which average spike-phase histograms across each SC depth deviate from a uniform circular distribution is shown as a bar plot to the right, where the color of each bar indicates the preferred phase at each depth. Note that spike-phase locking is strongest and occurs at the earliest phase in intermediate SC layers. From intermediate layers, both the strength of phase locking and phase lag decrease gradually with increasing distance in dorsal and ventral directions.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/1/7/e1500229/DC1

    Fig. S1. Population average power spectra of μECoG, SC, and intracortical recording sites.

    Fig. S2. SC current source density analysis.

    Fig. S3. Large-scale structure of envelope coupling modes.

    Fig. S4. Temporal variability of cortico-tectal high-frequency amplitude envelope correlation.

    Fig. S5. Cortical depth profile.

    Fig. S6. Effects of breathing on coordinated neural activity.

    Fig. S7. Cross-frequency coupling spectra and locally re-referenced signals.

    Fig. S8. Spatial maps of slow oscillation coherence.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Population average power spectra of μECoG, SC, and intracortical recording sites.
    • Fig. S2. SC current source density analysis.
    • Fig. S3. Large-scale structure of envelope coupling modes.
    • Fig. S4. Temporal variability of cortico-tectal high-frequency amplitude envelope correlation.
    • Fig. S5. Cortical depth profile.
    • Fig. S6. Effects of breathing on coordinated neural activity.
    • Fig. S7. Cross-frequency coupling spectra and locally re-referenced signals.
    • Fig. S8. Spatial maps of slow oscillation coherence.

    Download PDF

    Files in this Data Supplement: