Research ArticleNEUROSCIENCE

Rapid eye movement sleep promotes cortical plasticity in the developing brain

See allHide authors and affiliations

Science Advances  03 Jul 2015:
Vol. 1, no. 6, e1500105
DOI: 10.1126/sciadv.1500105
  • Fig. 1 REM sleep can be selectively deprived in developing cats.

    Sleep, n = 11; RSD, n = 12; NF, n = 10. Error bars, SEM. (A) Experimental design. We recorded baseline EEG/EMG signals and then performed MD during wake under lighted conditions. Then, we allowed animals 1 hour of sleep, RSD, or NF in complete darkness. (B) Time spent in each arousal state during the baseline, MD, and post-MD periods. RSD significantly reduced REM sleep. Both RSD and NF increased post-MD wake time. *P < 0.05 (wake); P < 0.05 (REM sleep); #P < 0.05 (NREM sleep), analysis of variance (ANOVA) on ranks followed by Dunn’s post hoc test versus sleep group. Sleep, n = 11 animals; RSD, n = 12; NF, n = 10. (C) RSD and NF decreased NREM bout duration. *P < 0.05, ANOVA on ranks followed by Dunn’s post hoc test. Sleep, n = 11 animals; RSD, n = 12; NF, n = 10. (D) EEG signals from V1 leads were Fourier-transformed, and the power spectrum during post-MD NREM sleep was normalized to baseline. Mean power in the 0.5- to 4-Hz and 4- to 10-Hz frequency bands did not differ between groups (ANOVA on ranks, P > 0.06; sleep, n = 11 animals; RSD, n = 11; NF, n = 10).

  • Fig. 2 RSD impairs ODP consolidation.

    (A) Representative optical maps from animals that received sleep, RSD, or NF after MD. Vascular maps show the pial surface of V1, cropped to remove out-of-focus areas and large vessels; scale bar, 1 mm. Angle and polar maps are color-coded to indicate the visual stimulus orientation that maximally drives the response at a given pixel. In polar maps, pixel brightness indicates the magnitude of the response. A polar map that is overall brighter in response to nondeprived than deprived eye stimulation indicates that an OD shift has occurred. This is also shown in the OD ratio maps, which show cortical areas dominated by stimulation of either eye. We quantified this shift by analyzing images pixel by pixel. (B) Responses at each pixel in the maps shown in (A) were binned into 70 OD categories on the basis of the ratio of their responsiveness to the nondeprived and deprived eyes (black line). The distribution was collapsed into seven OD categories to produce the classical seven-point scale of Hubel and Wiesel (3). (C) Average (±SEM) scalar measures of the OD shift. The OD shift after RSD was significantly reduced compared to those after sleep and NF, measured by the NBI (*P = 0.011, sleep versus RSD; P = 0.009, NF versus RSD), MI (*P = 0.003, sleep versus RSD; P = 0.013, NF versus RSD), and SI (*P < 0.001, sleep versus RSD; P = 0.034, NF versus RSD). One-way ANOVA followed by Fisher’s least significant difference (LSD) test (30); sleep, n = 9 hemispheres; RSD, n = 10; NF, n = 10. The hashed reference line represents comparable values from animals analyzed after 6 hours of MD only without subsequent sleep [adapted from (8)].

  • Fig. 3 RSD decreases ERK phosphorylation in V1.

    (A) Representative Western blots. pERK1, phospho-ERK; pCaMKIIα, phospho-CaMKIIα. (B) Average (±SEM) protein phosphorylation (displayed as percent of sleep group levels). Compared with sleep and NF, ERK1/2 phosphorylation was significantly reduced by RSD. *P < 0.05, sleep versus RSD; P < 0.05, NF versus RSD, ANOVA on ranks followed by Dunn’s post hoc test. Sleep, n = 12 hemispheres; RSD, n = 14; NF, n = 10.

  • Fig. 4 Patterns of V1 neuronal firing in REM sleep resemble patterns during MD.

    (A) Heat maps showing patterns of single V1 neuron activity (averaged in 10-s epochs and rank-transformed for display purposes) in a representative, freely moving cat (dark blue, firing rate minima; red, maxima). Data are contiguous 10-s segments of neuronal firing during REM sleep in the normal vision, baseline (pre-MD) period (left panel), MD in the awake animal (middle panel), and in the first hour of post-MD REM sleep (right panel). Each neuron is displayed along the y axis, time is displayed on the x axis, and intensity of unit firing is denoted by changes in color. Note how neuronal activity in REM sleep changes (from pre to post) to resemble activity during MD. (B) Cartoon shows a supervised learning algorithm [multilayer perceptron (MLP) neural network] trained to identify patterns of activity specific to the normal vision period (a) or MD period (b) (with 97 to 100% accuracy) and then used to screen pre-MD or post-MD REM sleep for matches to the MD pattern. The MLP architecture is composed of an input layer [x1-xi, corresponding to firing rates in individual neurons (N1-Ni) in the pre-MD waking or MD waking periods], a hidden processing layer, and an output layer. (C) Mean probability that neuronal activity in REM sleep matches activity during MD. Post-MD REM sleep data were divided into 2-hour segments and contained the average value for all bouts ≥1 min in length. An equivalent number of bouts (randomly selected) were used for the baseline (pre-MD) REM sleep comparison group. Mean probability of match to MD pattern was significantly higher in post-MD REM sleep than in baseline REM sleep (n = 5; *P < 0.006 versus baseline, F = 6.036, one-way ANOVA).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/1/6/e1500105/DC1

    Fig. S1. MD, serum cortisol levels, and timing of awakenings are similar across groups and magnitude of RSD effect comparable to pharmacological block of plasticity.

    Fig. S2. Single-unit recordings in V1 confirm the results from optical imaging.

    Fig. S3. RSD decreases cofilin levels, but not GluA1 phosphorylation, in V1.

    Fig. S4. RSD reduces ERK phosphorylation in nonvisual brain regions.

    Fig. S5. REM sleep in the first 2 hours after MD rescues and protects ODP from further disruption.

    Fig. S6. REM sleep is selectively reduced in the 2 hours after MD.

    Fig. S7. ERK phosphorylation is normal in the RSD + recovery and delayed RSD groups.

    Fig. S8. EEG activity in REM sleep correlates with ODP and ERK phosphorylation.

    Table S1. Numbers and ages of animals used in the 1-hour groups.

    Table S2. Numbers and ages of animals used in the 2-hour groups.

    Table S3. Training and cross-validation of an ANN using the MLP architecture.

    Table S4. Testing and performance of an ANN using the MLP architecture.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. MD, serum cortisol levels, and timing of awakenings are similar across groups and magnitude of RSD effect comparable to pharmacological block of plasticity.
    • Fig. S2. Single-unit recordings in V1 confirm the results from optical imaging.
    • Fig. S3. RSD decreases cofilin levels, but not GluA1 phosphorylation, in V1.
    • Fig. S4. RSD reduces ERK phosphorylation in nonvisual brain regions.
    • Fig. S5. REM sleep in the first 2 hours after MD rescues and protects ODP from further disruption.
    • Fig. S6. REM sleep is selectively reduced in the 2 hours after MD.
    • Fig. S7. ERK phosphorylation is normal in the RSD + recovery and delayed RSD groups.
    • Fig. S8. EEG activity in REM sleep correlates with ODP and ERK phosphorylation.
    • Table S1. Numbers and ages of animals used in the 1-hour groups.
    • Table S2. Numbers and ages of animals used in the 2-hour groups.
    • Table S3. Training and cross-validation of an ANN using the MLP architecture.
    • Table S4. Testing and performance of an ANN using the MLP architecture.

    Download PDF

    Files in this Data Supplement:

Navigate This Article