Synapse elimination and learning rules co-regulated by MHC class I H2-Db

Journal name:
Nature
Volume:
509,
Pages:
195–200
Date published:
DOI:
doi:10.1038/nature13154
Received
Accepted
Published online

Abstract

The formation of precise connections between retina and lateral geniculate nucleus (LGN) involves the activity-dependent elimination of some synapses, with strengthening and retention of others. Here we show that the major histocompatibility complex (MHC) class I molecule H2-Db is necessary and sufficient for synapse elimination in the retinogeniculate system. In mice lacking both H2-Kb and H2-Db (KbDb−/−), despite intact retinal activity and basal synaptic transmission, the developmentally regulated decrease in functional convergence of retinal ganglion cell synaptic inputs to LGN neurons fails and eye-specific layers do not form. Neuronal expression of just H2-Db in KbDb−/− mice rescues both synapse elimination and eye-specific segregation despite a compromised immune system. When patterns of stimulation mimicking endogenous retinal waves are used to probe synaptic learning rules at retinogeniculate synapses, long-term potentiation (LTP) is intact but long-term depression (LTD) is impaired in KbDb−/− mice. This change is due to an increase in Ca2+-permeable AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid) receptors. Restoring H2-Db to KbDb−/− neurons renders AMPA receptors Ca2+ impermeable and rescues LTD. These observations reveal an MHC-class-I-mediated link between developmental synapse pruning and balanced synaptic learning rules enabling both LTD and LTP, and demonstrate a direct requirement for H2-Db in functional and structural synapse pruning in CNS neurons.

At a glance

Figures

  1. Failure of retinogeniculate synapse elimination despite intact retinal waves in KbDb-/-.
    Figure 1: Failure of retinogeniculate synapse elimination despite intact retinal waves in KbDb−/−.

    ae, Impaired synapse elimination in KbDb−/− mice at P20–P24. a, Slice preparation used for whole-cell recording from dLGN neurons and stimulation of retinal ganglion cell (RGC) axons in the optic tract. The retinogeniculate projection is visualized by injecting CTB AF488 (green) into the contralateral eye. b, c, EPSC amplitude versus optic tract stimulus intensity. Insets: example traces. d, Cumulative probability histograms of single fibre synaptic strength (SF-AMPA). Inset: mean±s.e.m. for wild type (WT) (n = 12/N = 6); KbDb−/− (n = 23/N = 8), *P<0.05. e, Fibre fraction for wild type (n = 12/N = 6); KbDb−/− (n = 21/N = 8), **P<0.01, t-test for d, e. f, g, Intact retinal waves in KbDb−/− at P10–P12. f, Raster plots of single-unit spike trains recorded from 10 representative RGCs during retinal waves. g, Correlation indices versus inter-electrode distance for all cell pairs for wild type (N = 5) versus KbDb−/− (N = 6). Data correspond to mean values of medians from individual data sets and error bars represent s.e.m. n = cells/N = animals.

  2. H2-Db expression in neurons rescues synapse elimination and eye-specific segregation in KbDb-/- LGN.
    Figure 2: H2-Db expression in neurons rescues synapse elimination and eye-specific segregation in KbDb−/− LGN.

    ad, Rescue of synapse elimination at P20–P24. a, b, EPSC amplitudes versus optic tract stimulus intensity. Insets: example traces. c, Cumulative probability histogram of SF-AMPA. Inset: mean±s.e.m. for control: Db− (KbDb−/−;NSEDb, n = 19/N = 5). Rescue: Db+ (KbDb−/−;NSEDb+, n = 17/N = 7), **P<0.01. d, Fibre fraction is also rescued in KbDb−/−;NSEDb+ (n = 16/N = 7) compared to KbDb−/−;NSEDb (n = 18/N = 5), *P<0.05, Mann–Whitney U-test for c, d. Horizontal grey bars delineate Fig. 1e data (mean±s.e.m.). e, f, Rescue of eye-specific segregation in KbDb−/−;NSEDb+ at P34. e, Top: coronal sections of dLGN showing pattern of retinogeniculate projections from the ipsilateral (green) and contralateral (red) eyes. Bottom: region of ipsi-contra pixel (white) overlap between the two channels at 60% intensity threshold (T60%). f, Percentage of dLGN area occupied by ipsi-contra overlap. mean±s.e.m. for KbDb−/−;NSEDb (N = 3) and KbDb−/−;NSEDb+ (N = 4) (T60%) (*P<0.05, two way ANOVA) (see Extended Data Fig. 5). Horizontal grey bar indicates wild-type value at T60% (from ref. 14). n = cells/N = animals.

  3. Impaired LTD but intact LTP at retinogeniculate synapses in KbDb-/- induced with natural activity patterns.
    Figure 3: Impaired LTD but intact LTP at retinogeniculate synapses in KbDb−/− induced with natural activity patterns.

    a, Diagram illustrating basis for timing-dependent plasticity at developing retinogeniculate synapses. Spontaneous retinal waves propagate from ‘a, b’ towards ‘c, d’; neighbouring RGCs fire synchronously but asynchronously with respect to RGCs located elsewhere. Waves drive action potentials in postsynaptic LGN neurons with varying time delays between pre- and postsynaptic activity. Ages P8–13 were studied. b, c, Top: conditioning protocol for LTP (0ms latency; b) or LTD (1,100ms latency; c). Bottom: example membrane potential changes recorded in LGN neuron during conditioning protocol. OT, optic tract. df, Intact LTP in KbDb−/−. Single experiment showing LTP in wild type (d) and KbDb−/− (e). EPSC peak amplitude versus time. f, Summary of all 0ms latency experiments: EPSC peak amplitude (% change from baseline) versus time (n = 6/N = 6 for each; P>0.1, t-test). gi, Deficient LTD in KbDb−/−. Single experiment for wild type (g) and KbDb−/− (h). EPSC peak amplitude versus time. i, Summary of all 1,100ms latency experiments: EPSC peak amplitude (% change from baseline) versus time (n = 7/N = 7 for each; P<0.01, t-test). Grey bars indicate induction period. Insets: average EPSCs (30 traces) before (grey) and after (black) induction. f, i, 1min data binning. Ra, access resistance (MΩ). n = cells/N = animals.

  4. Increased Ca2+-permeable AMPA receptors at retinogeniculate synapses in KbDb-/- LGN.
    Figure 4: Increased Ca2+-permeable AMPA receptors at retinogeniculate synapses in KbDb−/− LGN.

    ad, Prolonged decay kinetics of IAMPA in KbDb−/− mice. a, Average IAMPA (5–10 EPSCs) for wild-type versus KbDb−/− LGN neurons. IAMPA half-width (ms) (b), IAMPA decay time (ms) (c) and peak amplitude (nA) (d) for wild type versus KbDb−/− (WT, n = 16/N = 4; KbDb−/−, n = 22/N = 5). e, Increased per cent inhibition of peak IAMPA by NASPM (100 µM) in KbDb−/− (n = 13/N = 4) versus wild type (n = 9/N = 3) (**P<0.01; NS, not significant; Mann–Whitney U-test for b, c). f, IAMPA IV curves (normalized to −40mV). g, Rectification index for wild type (n = 14/N = 3), KbDb−/− (n = 9/N = 3), KbDb−/− +20μM NASPM (n = 16/N = 4) or KbDb−/− +100μM NASPM (n = 6/N = 2) (***P<0.001 for wild type versus KbDb−/−; P>0.05 for wild type versus KbDb−/− +NASPM (20 or 100μM), Mann–Whitney U-test). Ages studied: P8–P13. Also, see Extended Data Fig. 7. n = cells/N = animals.

  5. Neuronal expression of H2-Db restores Ca2+-impermeable AMPA receptors and rescues LTD.
    Figure 5: Neuronal expression of H2-Db restores Ca2+-impermeable AMPA receptors and rescues LTD.

    ac, IAMPA half-width (ms) (a), IAMPA decay time (ms) (b), and peak amplitude (nA) (c) for KbDb−/−;NSEDb (n = 9/N = 2) and KbDb−/−;NSEDb+ (n = 11/N = 4). d, Reduced per cent inhibition of peak IAMPA by NASPM (100μM) in KbDb−/−;NSEDb+ (n = 10/N = 3) compared to KbDb−/−;NSEDb (n = 8/N = 2); *P<0.05, **P<0.01, ***P<0.001, Mann–Whitney U-test for ad. e, Rescue of IAMPA linear IV relationship in KbDb−/−;NSEDb+ LGN. Rectification index at +40 mV for KbDb−/−;NSEDb (n = 11/N = 3) and KbDb−/−;NSEDb+ (n = 13/N = 5) shows significant difference (**P<0.005); KbDb−/−;NSEDb+ (+NASPM) (n = 7/N = 3) is not significantly different to KbDb−/−;NSEDb+ (P>0.05), Mann–Whitney U-test. See also Extended Data Fig. 9. mean±s.e.m. f, g, LTD rescued in KbDb−/−;NSEDb+ LGN neurons. f, Ensemble average of all experiments at P8–9 (see Fig. 3). Grey bar indicates LTD induction period. 1min data binning. g, Average per cent change (mean±s.e.m.) for wild type (N = 7), KbDb−/− (N = 7), KbDb−/−;NSEDb(N = 7) and KbDb−/−;NSEDb+: (N = 9). *P<0.05, **P<0.01, NS, not significant, t-test. P8–13 ages studied. n = cells/N = animals.

  6. Comparison of retinogeniculate synaptic responses in wild type versus KbDb-/-.
    Extended Data Fig. 1: Comparison of retinogeniculate synaptic responses in wild type versus KbDb−/−.

    a, b, Examples of minimal stimulation for wild type (open circles) and KbDb−/− (filled circles). Plot of EPSC peak versus number of stimulations (grey box represents failures, >50%). c, No difference in onset latency of SF-AMPA between all genotypes. Onset latency of SF-AMPA was estimated using minimal stimulation as time (ms) to reach 10% of peak IAMPA from stimulation artefact (wild type: 3.0±0.3 (n = 12 cells/N = 6 animals); KbDb−/−: 2.7±0.1 (n = 23/N = 8); KbDb−/−;NSEDb+: 3.0±0.2 (n = 17/N = 7); KbDb−/−;NSEDb: 2.6±0.2 (n = 19/N = 5); P>0.5, t-test). d, Cumulative probability histogram shows no difference in Max-AMPA between wild type and KbDb−/−. Inset: mean±s.e.m. Wild type: 2.6±0.4nA (n = 14/N = 6); KbDb−/−: 2.9±0.4nA (n = 22/N = 8); P>0.1, Mann–Whitney U-test. eh, Presynaptic release probability at KbDb−/− retinogeniculate synapses is similar to wild type at P20–24. e, f, Examples of EPSCs evoked by paired-pulse stimulation of optic tract (20Hz) recorded in whole-cell mode in individual LGN neurons from wild type (e) versus KbDb−/− (f). g, h, Paired-pulse depression (PPD) (%) (EPSC 2/EPSC 1) over varying intervals. g, Wild type (open circle) versus KbDb−/− (filled circle) without cyclothiazide (CTZ), a blocker of AMPA receptor desensitization. Wild type versus KbDb−/−: 1Hz: 82.3±2.6 (n = 10) versus 77.4±3.4 (n = 8); 10Hz: 44.9±2.3 (n = 9) versus 42.8±3.9 (n = 9); 20Hz: 37. 1±2.1 (n = 10) versus 37.3±2.8 (n = 9) (P>0.1 for each). h, Wild type (open circle) versus KbDb−/− (filled circle) with CTZ (20μM). Wild type versus KbDb−/−: 1Hz: 79.0±2.4 (n = 9) versus 81.8±1.1 (n = 7); 10Hz: 59.2±3.0 (n = 8) versus 57.4±3.7 (n = 7); 20Hz: 58.5±2.3 (n = 8) versus 56.6±4.1 (n = 7) (P>0.1 for each). N = 4 for wild type; N = 3 for KbDb−/− for g, h. There was no significant difference in PPD between wild type and KbDb−/−, but note significant decrease of PPD +20μM CTZ versus 0μM CTZ application for both wild type and KbDb−/− at 10Hz and 20Hz (P<0.05). t-test. mean±s.e.m. n = cells/N = animals.

  7. Intact spatio-temporal pattern of retinal waves in KbDb-/- mice at P5-12.
    Extended Data Fig. 2: Intact spatio-temporal pattern of retinal waves in KbDb−/− mice at P5–12.

    a, Correlation indices as a function of inter-electrode distance for all cell pairs for wild type (grey) and KbDb−/− (black) at P5–P8. Data points correspond to mean values of medians from individual data sets and error bars represent s.e.m. be, Summary of temporal firing patterns for retinas isolated from wild-type (open squares) and KbDb−/− (filled squares) mice at P5–P8 (stage II, cholinergic waves) and P10–P12 (stage III, glutamatergic waves). Open circles correspond to the mean values of individual retinas. b, Interburst interval for stage II (in min): wild type: 1.9±0.2; KbDb−/−: 1.6±0.2; stage III: wild type: 1.0±0.3; KbDb−/−: 0.9±0.1. c, Firing rate during burst for stage II (in Hz): wild type: 8.6±1.9; KbDb−/−: 8.4±1.7; stage III: wild type: 19.7±2.1; KbDb−/−: 18.6±2.2. d, Burst duration for stage II (in seconds): wild type: 4.2±0.2; KbDb−/−: 3.8±0.1; stage III: wild type: 2.6±0.6; KbDb−/−: 2.1±0.2. e, Mean firing rate for stage II (in Hz): wild type: 0.5±0.1; KbDb−/−: 0.5±0.2; stage III: wild type: 1.0±0.1; KbDb−/−: 0.8±0.1, mean±s.e.m. (P>0.05 for each, t-test, N = 6 animals for each group, except when N = 5 for stage III wild type; non-blind experiments).

  8. Rescue of H2-Db expression in brain of KbDb-/-;NSEDb+ mice.
    Extended Data Fig. 3: Rescue of H2-Db expression in brain of KbDb−/−;NSEDb+ mice.

    a, Diagram of breeding strategy to generate KbDb−/−;NSEDb+ mice. KbDb−/− (white indicates absence of H2-Db) were crossed to NSEDb transgenic mice (black indicates presence of H2-Db in both body and brain). From F1 offspring, H2-Kb+/−H2-Db+/−;NSEDb+ mice (grey body, black brain) were selected and crossed to KbDb−/− mice further, generating KbDb−/−;NSEDb+ (black brain with white body indicates rescue of H2-Db expression in brain alone) and KbDb−/−;NSEDb littermate controls (white body, white brain). b, H2-Db-specific primers. Solid arrows: forward (exon 2) and reverse (exon 3) for ~200bp spliced mRNA as well as unspliced proRNA (~500 bp) (e; exon, i; intron). Dotted arrows: NSEDb-A (forward, NSE promoter region) and NSEDb-B1 and/or B2 (reverse, exon 2) for genotyping and mRNA detection20. c, RT–PCR showing rescue at P10 in thalamus of KbDb−/−;NSEDb+ mice cDNAs. Wild-type thalamus (WT(T)) shown as positive control; KbDb−/− thalamus (KbDb−/− (T)) and spleen (KbDb−/− (S)) as negative controls, and various organs from KbDb−/−;NSEDb+ mice (spleen (S), liver (L), gut (G), thalamus (T), hippocampus (H), cortex (C), retina (R)) were used as templates. d, Quantitative PCR comparing relative H2-Kb/H2-Db gene expression in wild type, KbDb−/− and KbDb−/−;NSEDb+ thalami. Left: results show small but highly significant rescue of H2-Db mRNA expression in KbDb−/−;NSEDb+ (KbDb−/−: 0.00003±0.00002; KbDb−/−;NSEDb+: 0.0035±0.00044 relative to wild type: 1.0056±0.032, ***P = 0.0001). Each point represents average relative gene expression for one mouse. Right: raw H2-Kb/H2-Db Ct values for each genotype (wild type: 27.4±0.1; KbDb−/−: 44.3±0.4; KbDb−/−;NSEDb+: 35.9±0.3; ***P<0.001), one-way ANOVA, mean±s.e.m. N = 5 animals for wild type, 7 for KbDb−/−, 10 for KbDb−/−;NSEDb+. e, Rescue of H2-Db protein in KbDb−/−;NSEDb+ brains at P60. Western blot of immunoprecipitation from whole brain (three pooled brains) lysate from KbDb−/− and KbDb−/−;NSEDb+. H2-Db-specific signal from KbDb−/−;NSEDb+ appears below IgG band.

  9. Cumulative probability distribution for SF-AMPA and Max-AMPA recorded at retinogeniculate synapses according to H2-Db genotype.
    Extended Data Fig. 4: Cumulative probability distribution for SF-AMPA and Max-AMPA recorded at retinogeniculate synapses according to H2-Db genotype.

    a, b, Max-AMPA (a) and SF-AMPA (b) showed similar cumulative probability histograms between wild type (black line), KbDb−/−;NSEDb+ (dashed grey line), KbDb−/− (dashed black line) and KbDb−/−;NSEDb (grey line). Number of experiments are the same as in the main text, except for Max-AMPA: for KbDb−/−;NSEDb: n = 18 cells/N = 5 animals; for KbDb−/−;NSEDb+: n = 16/N = 7 (P>0.05, Mann–Whitney U-test). Fibre fraction calculated from Max-AMPA and SF-AMPA measurements is similar between wild-type and KbDb−/−;NSEDb+ mice (Fig. 2d).

  10. Neuronal H2-Db expression in KbDb-/- mice rescues impaired eye-specific axonal segregation at P34.
    Extended Data Fig. 5: Neuronal H2-Db expression in KbDb−/− mice rescues impaired eye-specific axonal segregation at P34.

    a, Coronal sections of dLGN of KbDb−/−;NSEDb (top row) and KbDb−/−;NSEDb+ (rescue; bottom row) showing pattern of RGC axonal projections from the two eyes after intraocular tracer injections of CTB AF594 (red channel; contralateral eye injected) or AF488 (green channel; ipsilateral eye injected). Left: thresholded fluorescent images of dLGN at 60% maximum signal intensity (see Fig. 2e). Right: overlap of RGC projections (white pixels) from ipsilateral and contralateral eyes displayed for 20%, 40% and 60% maximal threshold for KbDb−/−;NSEDb (top) and KbDb−/−;NSEDb+ (bottom). Overlap = pixels labelled in both red and green channels. b, Mean percentage dLGN area±s.e.m. pixel overlap for KbDb−/−;NSEDb (filled squares; N = 3) versus KbDb−/−;NSEDb+ (open triangles; N = 4): 0% threshold: 8.3±1.1 versus 1.2±0.3; 20% threshold: 12.2±1.5 versus 2.9±0.7; 40% threshold: 17.9±1.8 versus 5.3±0.9; 60% threshold: 21.1±1.8 versus 7.2±0.9; 80% threshold: 23.2±1.3 versus 11.1±1.2; 100%: 25.3±1.9 versus 14.0±1.5 (P<0.05, two-way ANOVA). See Methods and ref. 47.

  11. Intact release probability at KbDb-/- retinogeniculate synapses before eye opening.
    Extended Data Fig. 6: Intact release probability at KbDb−/− retinogeniculate synapses before eye opening.

    Paired-pulse stimulation was delivered to the optic tract at 10Hz, similar to the natural firing frequency of RGCs (Extended Data Fig. 2c), and whole-cell recordings were made from LGN neurons in slices aged between P8–13. Paired-pulse stimulation resulted in synaptic depression, represented as EPSC 2 divided by EPSC 1 (%). In 0μM CTZ (left panel): wild type: 67.0±2.9 (n = 11/N = 4); KbDb−/−: 64.2±2.8 (n = 7/N = 2). In 20μM CTZ (right panel): wild type: 48.6±4.9 (n = 8/N = 3); KbDb−/−: 46.6±2.5 (n = 7/N = 2) (P>0.1 for each, t-test), mean±s.e.m. 20mM BAPTA containing Cs+-internal solution was used for this experiment due to prolonged kinetics of EPSCs in KbDb−/−. The identical paired-pulse ratios between wild type and KbDb−/− are consistent with the conclusion that presynaptic release probability is intact at P8–13 retinogeniculate synapses in KbDb−/− mice. (See also Extended Data Fig. 1e–h for similar conclusion at P20–24, after synapse elimination is largely complete.)

  12. Normal NMDA/AMPA ratio but increased Ca2+-permeable AMPA receptors at retinogeniculate synapses in KbDb-/- mice.
    Extended Data Fig. 7: Normal NMDA/AMPA ratio but increased Ca2+-permeable AMPA receptors at retinogeniculate synapses in KbDb−/− mice.

    a, b, NMDA/AMPA ratio is unchanged in KbDb−/− mice. a, NMDA/AMPA ratio (%): peak IAMPA measured at −70mV (+20μM SR95531) versus peak INMDA at +40mV (+20μM SR95531 + 20μM DNQX): wild type: 61±6.8 (n = 10/N = 4); KbDb−/−: 70.6 ±12.1 (n = 7/N = 3) (P>0.1, t-test) mean±s.e.m. b, Example recordings from individual neurons for wild type (left) and KbDb−/− (right). APV (100μM) was added at the end of each experiment to confirm NMDA-mediated synaptic currents. D600 in pipette. c, Example showing effect of NASPM (100μM bath) on IAMPA: note significant blockade of IAMPA in KbDb−/−. Grey line, before NASPM; black line, after NASPM (5 traces averaged for single cell). SR95531 (20μM) added to bath for ac. d, Examples for IAMPA normalized to EPSC amplitude at −40mV. Note reduction in EPSC amplitude at +40mV in KbDb−/− but not wild type. 100μM APV + 20μM SR95531 in bath. Spermine (100μM) and D600 (100μM) in pipette. Ages: P8–13. Experimenter was aware of genotype due to obvious differences in time course of EPSCs and effects of NASPM. e, Example western blot (left) and GluR1/GluR2 ratio (right) of P22 thalamus; wild type: 1.0 ±0.1 (N = 12); KbDb−/−: 1.3±0.1 (N = 13) (P = 0.07). f, Example western blot (left) and GluR1/GluR2 ratio (right) of cultured cortical neurons; wild type: 1.0 ±0.1 (N = 4); KbDb−/−: 2.3±0.7 (N = 4) (*P = 0.03). Mann–Whitney U-test for e, f, n = cells/N = animals.

  13. NASPM-dependent rescue of LTD in KbDb-/- LGN at P8-13.
    Extended Data Fig. 8: NASPM-dependent rescue of LTD in KbDb−/− LGN at P8–13.

    a, Summary of all 1,100ms latency experiments: EPSC peak amplitude (% change from baseline) versus time (n = 7/N = 7 for 0μM NASPM; n = 5/N = 5 for 20μM NASPM). Grey bar = induction period (Methods). 1min data binning. b, Average of per cent change (mean±s.e.m.); KbDb−/− 0μM NASPM: 105±8.2 (N = 7); KbDb−/− +20μM NASPM: 72.5±2.2 (N = 5). **P<0.01, t-test, n = cells/N = animals.

  14. Neuronal H2-Db expression decreases Ca2+ permeability of AMPA receptors at retinogeniculate synapses in KbDb-/-;NSEDb+ mice.
    Extended Data Fig. 9: Neuronal H2-Db expression decreases Ca2+ permeability of AMPA receptors at retinogeniculate synapses in KbDb−/−;NSEDb+ mice.

    a, NASPM blockade of IAMPA is significantly reduced in KbDb−/−;NSEDb+ mice. Example trace of NASPM effect on IAMPA recorded from KbDb−/−;NSEDb (left) or KbDb−/−;NSEDb+ (right) individual LGN neuron. Grey line, before NASPM; black line, after NASPM application (5 traces averaged for single cell); 20 µM SR95531 in bath. b, Internal spermine-dependent block of IAMPA at positive membrane potentials is rescued in KbDb−/−;NSEDb+ LGN neurons. Example recordings for IAMPA normalized to EPSC amplitude at −40mV from individual neurons. Note reduction in EPSC amplitude at +40mV in KbDb−/−; NSEDb, but restored to wild-type level in KbDb−/−;NSEDb+. 100μM APV + 20μM SR95531 in bath. Spermine (100μM) and D600 (100μM) in internal solution.

References

  1. Meister, M., Wong, R. O., Baylor, D. A. & Shatz, C. J. Synchronous bursts of action potentials in ganglion cells of the developing mammalian retina. Science 252, 939943 (1991)
  2. Feller, M. B., Wellis, D. P., Stellwagen, D., Werblin, F. S. & Shatz, C. J. Requirement for cholinergic synaptic transmission in the propagation of spontaneous retinal waves. Science 272, 11821187 (1996)
  3. Wong, R. O., Meister, M. & Shatz, C. J. Transient period of correlated bursting activity during development of the mammalian retina. Neuron 11, 923938 (1993)
  4. Mooney, R., Penn, A. A., Gallego, R. & Shatz, C. J. Thalamic relay of spontaneous retinal activity prior to vision. Neuron 17, 863874 (1996)
  5. Weliky, M. & Katz, L. C. Correlational structure of spontaneous neuronal activity in the developing lateral geniculate nucleus in vivo. Science 285, 599604 (1999)
  6. Ackman, J. B., Burbridge, T. J. & Crair, M. C. Retinal waves coordinate patterned activity throughout the developing visual system. Nature 490, 219225 (2012)
  7. Penn, A. A., Riquelme, P. A., Feller, M. B. & Shatz, C. J. Competition in retinogeniculate patterning driven by spontaneous activity. Science 279, 21082112 (1998)
  8. Huberman, A. D., Feller, M. B. & Chapman, B. Mechanisms underlying development of visual maps and receptive fields. Annu. Rev. Neurosci. 31, 479509 (2008)
  9. Stevens, B. et al. The classical complement cascade mediates CNS synapse elimination. Cell 131, 11641178 (2007)
  10. Huh, G. S. et al. Functional requirement for class I MHC in CNS development and plasticity. Science 290, 21552159 (2000)
  11. Bjartmar, L. et al. Neuronal pentraxins mediate synaptic refinement in the developing visual system. J. Neurosci. 26, 62696281 (2006)
  12. Corriveau, R. A., Huh, G. S. & Shatz, C. J. Regulation of class I MHC gene expression in the developing and mature CNS by neural activity. Neuron 21, 505520 (1998)
  13. Glynn, M. W. et al. MHCI negatively regulates synapse density during the establishment of cortical connections. Nature Neurosci. 14, 442451 (2011)
  14. Datwani, A. et al. Classical MHCI molecules regulate retinogeniculate refinement and limit ocular dominance plasticity. Neuron 64, 463470 (2009)
  15. Chen, C. & Regehr, W. G. Developmental remodeling of the retinogeniculate synapse. Neuron 28, 955966 (2000)
  16. Hooks, B. M. & Chen, C. Distinct roles for spontaneous and visual activity in remodeling of the retinogeniculate synapse. Neuron 52, 281291 (2006)
  17. Stevens, C. F. & Wang, Y. Changes in reliability of synaptic function as a mechanism for plasticity. Nature 371, 704707 (1994)
  18. Torborg, C. L., Hansen, K. A. & Feller, M. B. High frequency, synchronized bursting drives eye-specific segregation of retinogeniculate projections. Nature Neurosci. 8, 7278 (2005)
  19. Vugmeyster, Y. et al. Major histocompatibility complex (MHC) class I KbDb−/− deficient mice possess functional CD8+ T cells and natural killer cells. Proc. Natl Acad. Sci. USA 95, 1249212497 (1998)
  20. Rall, G. F., Mucke, L. & Oldstone, M. B. Consequences of cytotoxic T lymphocyte interaction with major histocompatibility complex class I-expressing neurons in vivo. J. Exp. Med. 182, 12011212 (1995)
  21. Shatz, C. J. & Kirkwood, P. A. Prenatal development of functional connections in the cat's retinogeniculate pathway. J. Neurosci. 4, 13781397 (1984)
  22. Shatz, C. J. Emergence of order in visual system development. Proc. Natl Acad. Sci. USA 93, 602608 (1996)
  23. Torborg, C. L. & Feller, M. B. Unbiased analysis of bulk axonal segregation patterns. J. Neurosci. Methods 135, 1726 (2004)
  24. Zhou, Q., Homma, K. J. & Poo, M. M. Shrinkage of dendritic spines associated with long-term depression of hippocampal synapses. Neuron 44, 749757 (2004)
  25. Bastrikova, N., Gardner, G. A., Reece, J. M., Jeromin, A. & Dudek, S. M. Synapse elimination accompanies functional plasticity in hippocampal neurons. Proc. Natl Acad. Sci. USA 105, 31233127 (2008)
  26. Malenka, R. C. & Bear, M. F. LTP and LTD: an embarrassment of riches. Neuron 44, 521 (2004)
  27. Yuste, R. & Bonhoeffer, T. Morphological changes in dendritic spines associated with long-term synaptic plasticity. Annu. Rev. Neurosci. 24, 10711089 (2001)
  28. Mu, Y. & Poo, M. M. Spike timing-dependent LTP/LTD mediates visual experience-dependent plasticity in a developing retinotectal system. Neuron 50, 115125 (2006)
  29. Mooney, R., Madison, D. V. & Shatz, C. J. Enhancement of transmission at the developing retinogeniculate synapse. Neuron 10, 815825 (1993)
  30. Ziburkus, J., Dilger, E. K., Lo, F. S. & Guido, W. LTD and LTP at the developing retinogeniculate synapse. J. Neurophysiol. 102, 30823090 (2009)
  31. Butts, D. A., Kanold, P. O. & Shatz, C. J. A burst-based “Hebbian” learning rule at retinogeniculate synapses links retinal waves to activity-dependent refinement. PLoS Biol. 5, e61 (2007)
  32. Shah, R. D. & Crair, M. C. Retinocollicular synapse maturation and plasticity are regulated by correlated retinal waves. J. Neurosci. 28, 292303 (2008)
  33. Zhang, J., Ackman, J. B., Xu, H. P. & Crair, M. C. Visual map development depends on the temporal pattern of binocular activity in mice. Nature Neurosci. 15, 298307 (2011)
  34. Liu, S. Q. & Cull-Candy, S. G. Synaptic activity at calcium-permeable AMPA receptors induces a switch in receptor subtype. Nature 405, 454458 (2000)
  35. Cull-Candy, S., Kelly, L. & Farrant, M. Regulation of Ca2+-permeable AMPA receptors: synaptic plasticity and beyond. Curr. Opin. Neurobiol. 16, 288297 (2006)
  36. Isaac, J. T., Ashby, M. C. & McBain, C. J. The role of the GluR2 subunit in AMPA receptor function and synaptic plasticity. Neuron 54, 859871 (2007)
  37. Goel, A. et al. Cross-modal regulation of synaptic AMPA receptors in primary sensory cortices by visual experience. Nature Neurosci. 9, 10011003 (2006)
  38. Hohnke, C. D., Oray, S. & Sur, M. Activity-dependent patterning of retinogeniculate axons proceeds with a constant contribution from AMPA and NMDA receptors. J. Neurosci. 20, 80518060 (2000)
  39. Jia, Z. et al. Enhanced LTP in mice deficient in the AMPA receptor GluR2. Neuron 17, 945956 (1996)
  40. Toyoda, H. et al. Long-term depression requires postsynaptic AMPA GluR2 receptor in adult mouse cingulate cortex. J. Cell. Physiol. 211, 336343 (2007)
  41. Needleman, L. A., Liu, X. B., El-Sabeawy, F., Jones, E. G. & McAllister, A. K. MHC class I molecules are present both pre- and postsynaptically in the visual cortex during postnatal development and in adulthood. Proc. Natl Acad. Sci. USA 107, 1699917004 (2010)
  42. Stefansson, H. et al. Common variants conferring risk of schizophrenia. Nature 460, 744747 (2009)
  43. Ripke, S. et al. Genome-wide association analysis identifies 13 new risk loci for schizophrenia. Nature Genet. 45, 11501159 (2013)
  44. Turner, J. P. & Salt, T. E. Characterization of sensory and corticothalamic excitatory inputs to rat thalamocortical neurones in vitro. J. Physiol. (Lond.) 510, 829843 (1998)
  45. Rae, J., Cooper, K., Gates, P. & Watsky, M. Low access resistance perforated patch recordings using amphotericin B. J. Neurosci. Methods 37, 1526 (1991)
  46. Torborg, C. L., Hansen, K. A. & Feller, M. B. High frequency, synchronized bursting drives eye-specific segregation of retinogeniculate projections. Nature Neurosci. 8, 7278 (2005)
  47. Torborg, C. L. & Feller, M. B. Unbiased analysis of bulk axonal segregation patterns. J. Neurosci. Methods 135, 1726 (2004)
  48. Adelson, J. D. et al. Neuroprotection from stroke in the absence of MHCI or PirB. Neuron 73, 11001107 (2012)
  49. Johnson, M. W., Chotiner, J. K. & Watson, J. B. Isolation and characterization of synaptoneurosomes from single rat hippocampal slices. J. Neurosci. Methods 77, 151156 (1997)
  50. Yin, Y., Edelman, G. M. & Vanderklish, P. W. The brain-derived neurotrophic factor enhances synthesis of Arc in synaptoneurosomes. Proc. Natl Acad. Sci. USA 99, 23682373 (2002)
  51. Viesselmann, C., Ballweg, J., Lumbard, D. & Dent, E. W. Nucleofection and primary culture of embryonic mouse hippocampal and cortical neurons. J. Vis. Exp. (2011)

Download references

Author information

Affiliations

  1. Departments of Biology and Neurobiology and Bio-X, James H. Clark Center, 318 Campus Drive, Stanford, California 94305, USA

    • Hanmi Lee,
    • Barbara K. Brott,
    • Jaimie D. Adelson,
    • Sarah Cheng,
    • Akash Datwani &
    • Carla J. Shatz
  2. Department of Molecular and Cell Biology & Helen Wills Neuroscience Institute, University of California, Berkeley, California 94720, USA

    • Lowry A. Kirkby &
    • Marla B. Feller
  3. Present address: Sage Bionetworks, 1100 Fairview Avenue N., Seattle, Washington 98109, USA.

    • Akash Datwani

Contributions

H.L. and C.J.S. designed all experiments, analysed and reviewed all results and wrote manuscript. Data contributions are as follows: electrophysiology experiment by H.L.; multi-electrode array experiments by L.A.K. and M.B.F. B.K.B. designed H2-Db monoclonal antibody and performed western blots. H.L. designed and performed RT–PCR experiments. A.D. performed RGC neuronal tract tracing experiments and analysis. J.D.A. and S.C. performed Taqman qPCR.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Extended data figures and tables

Extended Data Figures

  1. Extended Data Figure 1: Comparison of retinogeniculate synaptic responses in wild type versus KbDb−/−. (252 KB)

    a, b, Examples of minimal stimulation for wild type (open circles) and KbDb−/− (filled circles). Plot of EPSC peak versus number of stimulations (grey box represents failures, >50%). c, No difference in onset latency of SF-AMPA between all genotypes. Onset latency of SF-AMPA was estimated using minimal stimulation as time (ms) to reach 10% of peak IAMPA from stimulation artefact (wild type: 3.0±0.3 (n = 12 cells/N = 6 animals); KbDb−/−: 2.7±0.1 (n = 23/N = 8); KbDb−/−;NSEDb+: 3.0±0.2 (n = 17/N = 7); KbDb−/−;NSEDb: 2.6±0.2 (n = 19/N = 5); P>0.5, t-test). d, Cumulative probability histogram shows no difference in Max-AMPA between wild type and KbDb−/−. Inset: mean±s.e.m. Wild type: 2.6±0.4nA (n = 14/N = 6); KbDb−/−: 2.9±0.4nA (n = 22/N = 8); P>0.1, Mann–Whitney U-test. eh, Presynaptic release probability at KbDb−/− retinogeniculate synapses is similar to wild type at P20–24. e, f, Examples of EPSCs evoked by paired-pulse stimulation of optic tract (20Hz) recorded in whole-cell mode in individual LGN neurons from wild type (e) versus KbDb−/− (f). g, h, Paired-pulse depression (PPD) (%) (EPSC 2/EPSC 1) over varying intervals. g, Wild type (open circle) versus KbDb−/− (filled circle) without cyclothiazide (CTZ), a blocker of AMPA receptor desensitization. Wild type versus KbDb−/−: 1Hz: 82.3±2.6 (n = 10) versus 77.4±3.4 (n = 8); 10Hz: 44.9±2.3 (n = 9) versus 42.8±3.9 (n = 9); 20Hz: 37. 1±2.1 (n = 10) versus 37.3±2.8 (n = 9) (P>0.1 for each). h, Wild type (open circle) versus KbDb−/− (filled circle) with CTZ (20μM). Wild type versus KbDb−/−: 1Hz: 79.0±2.4 (n = 9) versus 81.8±1.1 (n = 7); 10Hz: 59.2±3.0 (n = 8) versus 57.4±3.7 (n = 7); 20Hz: 58.5±2.3 (n = 8) versus 56.6±4.1 (n = 7) (P>0.1 for each). N = 4 for wild type; N = 3 for KbDb−/− for g, h. There was no significant difference in PPD between wild type and KbDb−/−, but note significant decrease of PPD +20μM CTZ versus 0μM CTZ application for both wild type and KbDb−/− at 10Hz and 20Hz (P<0.05). t-test. mean±s.e.m. n = cells/N = animals.

  2. Extended Data Figure 2: Intact spatio-temporal pattern of retinal waves in KbDb−/− mice at P5–12. (117 KB)

    a, Correlation indices as a function of inter-electrode distance for all cell pairs for wild type (grey) and KbDb−/− (black) at P5–P8. Data points correspond to mean values of medians from individual data sets and error bars represent s.e.m. be, Summary of temporal firing patterns for retinas isolated from wild-type (open squares) and KbDb−/− (filled squares) mice at P5–P8 (stage II, cholinergic waves) and P10–P12 (stage III, glutamatergic waves). Open circles correspond to the mean values of individual retinas. b, Interburst interval for stage II (in min): wild type: 1.9±0.2; KbDb−/−: 1.6±0.2; stage III: wild type: 1.0±0.3; KbDb−/−: 0.9±0.1. c, Firing rate during burst for stage II (in Hz): wild type: 8.6±1.9; KbDb−/−: 8.4±1.7; stage III: wild type: 19.7±2.1; KbDb−/−: 18.6±2.2. d, Burst duration for stage II (in seconds): wild type: 4.2±0.2; KbDb−/−: 3.8±0.1; stage III: wild type: 2.6±0.6; KbDb−/−: 2.1±0.2. e, Mean firing rate for stage II (in Hz): wild type: 0.5±0.1; KbDb−/−: 0.5±0.2; stage III: wild type: 1.0±0.1; KbDb−/−: 0.8±0.1, mean±s.e.m. (P>0.05 for each, t-test, N = 6 animals for each group, except when N = 5 for stage III wild type; non-blind experiments).

  3. Extended Data Figure 3: Rescue of H2-Db expression in brain of KbDb−/−;NSEDb+ mice. (181 KB)

    a, Diagram of breeding strategy to generate KbDb−/−;NSEDb+ mice. KbDb−/− (white indicates absence of H2-Db) were crossed to NSEDb transgenic mice (black indicates presence of H2-Db in both body and brain). From F1 offspring, H2-Kb+/−H2-Db+/−;NSEDb+ mice (grey body, black brain) were selected and crossed to KbDb−/− mice further, generating KbDb−/−;NSEDb+ (black brain with white body indicates rescue of H2-Db expression in brain alone) and KbDb−/−;NSEDb littermate controls (white body, white brain). b, H2-Db-specific primers. Solid arrows: forward (exon 2) and reverse (exon 3) for ~200bp spliced mRNA as well as unspliced proRNA (~500 bp) (e; exon, i; intron). Dotted arrows: NSEDb-A (forward, NSE promoter region) and NSEDb-B1 and/or B2 (reverse, exon 2) for genotyping and mRNA detection20. c, RT–PCR showing rescue at P10 in thalamus of KbDb−/−;NSEDb+ mice cDNAs. Wild-type thalamus (WT(T)) shown as positive control; KbDb−/− thalamus (KbDb−/− (T)) and spleen (KbDb−/− (S)) as negative controls, and various organs from KbDb−/−;NSEDb+ mice (spleen (S), liver (L), gut (G), thalamus (T), hippocampus (H), cortex (C), retina (R)) were used as templates. d, Quantitative PCR comparing relative H2-Kb/H2-Db gene expression in wild type, KbDb−/− and KbDb−/−;NSEDb+ thalami. Left: results show small but highly significant rescue of H2-Db mRNA expression in KbDb−/−;NSEDb+ (KbDb−/−: 0.00003±0.00002; KbDb−/−;NSEDb+: 0.0035±0.00044 relative to wild type: 1.0056±0.032, ***P = 0.0001). Each point represents average relative gene expression for one mouse. Right: raw H2-Kb/H2-Db Ct values for each genotype (wild type: 27.4±0.1; KbDb−/−: 44.3±0.4; KbDb−/−;NSEDb+: 35.9±0.3; ***P<0.001), one-way ANOVA, mean±s.e.m. N = 5 animals for wild type, 7 for KbDb−/−, 10 for KbDb−/−;NSEDb+. e, Rescue of H2-Db protein in KbDb−/−;NSEDb+ brains at P60. Western blot of immunoprecipitation from whole brain (three pooled brains) lysate from KbDb−/− and KbDb−/−;NSEDb+. H2-Db-specific signal from KbDb−/−;NSEDb+ appears below IgG band.

  4. Extended Data Figure 4: Cumulative probability distribution for SF-AMPA and Max-AMPA recorded at retinogeniculate synapses according to H2-Db genotype. (111 KB)

    a, b, Max-AMPA (a) and SF-AMPA (b) showed similar cumulative probability histograms between wild type (black line), KbDb−/−;NSEDb+ (dashed grey line), KbDb−/− (dashed black line) and KbDb−/−;NSEDb (grey line). Number of experiments are the same as in the main text, except for Max-AMPA: for KbDb−/−;NSEDb: n = 18 cells/N = 5 animals; for KbDb−/−;NSEDb+: n = 16/N = 7 (P>0.05, Mann–Whitney U-test). Fibre fraction calculated from Max-AMPA and SF-AMPA measurements is similar between wild-type and KbDb−/−;NSEDb+ mice (Fig. 2d).

  5. Extended Data Figure 5: Neuronal H2-Db expression in KbDb−/− mice rescues impaired eye-specific axonal segregation at P34. (462 KB)

    a, Coronal sections of dLGN of KbDb−/−;NSEDb (top row) and KbDb−/−;NSEDb+ (rescue; bottom row) showing pattern of RGC axonal projections from the two eyes after intraocular tracer injections of CTB AF594 (red channel; contralateral eye injected) or AF488 (green channel; ipsilateral eye injected). Left: thresholded fluorescent images of dLGN at 60% maximum signal intensity (see Fig. 2e). Right: overlap of RGC projections (white pixels) from ipsilateral and contralateral eyes displayed for 20%, 40% and 60% maximal threshold for KbDb−/−;NSEDb (top) and KbDb−/−;NSEDb+ (bottom). Overlap = pixels labelled in both red and green channels. b, Mean percentage dLGN area±s.e.m. pixel overlap for KbDb−/−;NSEDb (filled squares; N = 3) versus KbDb−/−;NSEDb+ (open triangles; N = 4): 0% threshold: 8.3±1.1 versus 1.2±0.3; 20% threshold: 12.2±1.5 versus 2.9±0.7; 40% threshold: 17.9±1.8 versus 5.3±0.9; 60% threshold: 21.1±1.8 versus 7.2±0.9; 80% threshold: 23.2±1.3 versus 11.1±1.2; 100%: 25.3±1.9 versus 14.0±1.5 (P<0.05, two-way ANOVA). See Methods and ref. 47.

  6. Extended Data Figure 6: Intact release probability at KbDb−/− retinogeniculate synapses before eye opening. (66 KB)

    Paired-pulse stimulation was delivered to the optic tract at 10Hz, similar to the natural firing frequency of RGCs (Extended Data Fig. 2c), and whole-cell recordings were made from LGN neurons in slices aged between P8–13. Paired-pulse stimulation resulted in synaptic depression, represented as EPSC 2 divided by EPSC 1 (%). In 0μM CTZ (left panel): wild type: 67.0±2.9 (n = 11/N = 4); KbDb−/−: 64.2±2.8 (n = 7/N = 2). In 20μM CTZ (right panel): wild type: 48.6±4.9 (n = 8/N = 3); KbDb−/−: 46.6±2.5 (n = 7/N = 2) (P>0.1 for each, t-test), mean±s.e.m. 20mM BAPTA containing Cs+-internal solution was used for this experiment due to prolonged kinetics of EPSCs in KbDb−/−. The identical paired-pulse ratios between wild type and KbDb−/− are consistent with the conclusion that presynaptic release probability is intact at P8–13 retinogeniculate synapses in KbDb−/− mice. (See also Extended Data Fig. 1e–h for similar conclusion at P20–24, after synapse elimination is largely complete.)

  7. Extended Data Figure 7: Normal NMDA/AMPA ratio but increased Ca2+-permeable AMPA receptors at retinogeniculate synapses in KbDb−/− mice. (210 KB)

    a, b, NMDA/AMPA ratio is unchanged in KbDb−/− mice. a, NMDA/AMPA ratio (%): peak IAMPA measured at −70mV (+20μM SR95531) versus peak INMDA at +40mV (+20μM SR95531 + 20μM DNQX): wild type: 61±6.8 (n = 10/N = 4); KbDb−/−: 70.6 ±12.1 (n = 7/N = 3) (P>0.1, t-test) mean±s.e.m. b, Example recordings from individual neurons for wild type (left) and KbDb−/− (right). APV (100μM) was added at the end of each experiment to confirm NMDA-mediated synaptic currents. D600 in pipette. c, Example showing effect of NASPM (100μM bath) on IAMPA: note significant blockade of IAMPA in KbDb−/−. Grey line, before NASPM; black line, after NASPM (5 traces averaged for single cell). SR95531 (20μM) added to bath for ac. d, Examples for IAMPA normalized to EPSC amplitude at −40mV. Note reduction in EPSC amplitude at +40mV in KbDb−/− but not wild type. 100μM APV + 20μM SR95531 in bath. Spermine (100μM) and D600 (100μM) in pipette. Ages: P8–13. Experimenter was aware of genotype due to obvious differences in time course of EPSCs and effects of NASPM. e, Example western blot (left) and GluR1/GluR2 ratio (right) of P22 thalamus; wild type: 1.0 ±0.1 (N = 12); KbDb−/−: 1.3±0.1 (N = 13) (P = 0.07). f, Example western blot (left) and GluR1/GluR2 ratio (right) of cultured cortical neurons; wild type: 1.0 ±0.1 (N = 4); KbDb−/−: 2.3±0.7 (N = 4) (*P = 0.03). Mann–Whitney U-test for e, f, n = cells/N = animals.

  8. Extended Data Figure 8: NASPM-dependent rescue of LTD in KbDb−/− LGN at P8–13. (109 KB)

    a, Summary of all 1,100ms latency experiments: EPSC peak amplitude (% change from baseline) versus time (n = 7/N = 7 for 0μM NASPM; n = 5/N = 5 for 20μM NASPM). Grey bar = induction period (Methods). 1min data binning. b, Average of per cent change (mean±s.e.m.); KbDb−/− 0μM NASPM: 105±8.2 (N = 7); KbDb−/− +20μM NASPM: 72.5±2.2 (N = 5). **P<0.01, t-test, n = cells/N = animals.

  9. Extended Data Figure 9: Neuronal H2-Db expression decreases Ca2+ permeability of AMPA receptors at retinogeniculate synapses in KbDb−/−;NSEDb+ mice. (130 KB)

    a, NASPM blockade of IAMPA is significantly reduced in KbDb−/−;NSEDb+ mice. Example trace of NASPM effect on IAMPA recorded from KbDb−/−;NSEDb (left) or KbDb−/−;NSEDb+ (right) individual LGN neuron. Grey line, before NASPM; black line, after NASPM application (5 traces averaged for single cell); 20 µM SR95531 in bath. b, Internal spermine-dependent block of IAMPA at positive membrane potentials is rescued in KbDb−/−;NSEDb+ LGN neurons. Example recordings for IAMPA normalized to EPSC amplitude at −40mV from individual neurons. Note reduction in EPSC amplitude at +40mV in KbDb−/−; NSEDb, but restored to wild-type level in KbDb−/−;NSEDb+. 100μM APV + 20μM SR95531 in bath. Spermine (100μM) and D600 (100μM) in internal solution.

Additional data