Research ArticleHUMAN GENETICS

Disruptive variants of CSDE1 associate with autism and interfere with neuronal development and synaptic transmission

See allHide authors and affiliations

Science Advances  25 Sep 2019:
Vol. 5, no. 9, eaax2166
DOI: 10.1126/sciadv.aax2166
  • Fig. 1 Spectrum of CSDE1 LGD variants and patient facial features.

    (A) Diagram of the canonical CSDE1 isoform (NM_001242891.1 and NP_001229820.1). The locations of LGD variants are indicated. (B) Pedigrees of eight families with de novo LGD variants (above) and eight families with transmitted LGD variants (below). Carrier parents or sibling in at least four families (PU2, BU2, SU1, and PU1) are affected or show substantial family history.

  • Fig. 2 Enrichment analyses of Csde1 RNA binding targets.

    (A) Bar plot shows Csde1-binding targets significantly enriched in seven ASD-related gene sets, especially the FMRP RNA binding targets. (B) Venn diagram shows the overlap of the RNA binding targets of CSDE1 and other ASD-associated RBPs (FMRP and RBFOX). (C) Venn diagram shows the overlap of the SFARI genes that are RNA binding targets of CSDE1, FMRP, and RBFOX, with gene names indicated. (D) Bar plot shows Csde1-binding targets enriched in neuronal development– and synapse development–related cell components [Fisher’s exact test, false discovery rate (FDR)–corrected]. Top 15 significant cell components are shown.

  • Fig. 3 Disruption of Csde1 interferes with neuronal development.

    (A) Csde1 KD promoted neurite and axon growth (NC, 73 neurons; Csde1 KD1, 60 neurons; Csde1 KD2, 53 neurons; Csde1 KD1 + WT, 82 neurons). Neurons were colabeled with 4′,6-diamidino-2-phenylindole (DAPI) (nuclei), green fluorescent protein (GFP) (overall neuronal morphology), and SMI 312 (axon). Scale bar, 100 μm. (B) Csde1 KD reduced the complexity of dendritic arborization (NC, 32 neurons; Csde1 KD1, 21 neurons; Csde1 KD2, 42 neurons; Csde1 KD2 + WT, 27 neurons; Csde1 KD2 + Ctnnb1, 26 neurons). (C) Csde1 KD disrupted dendritic spine morphogenesis and maturation (NC, 27 neurons; Csde1 KD1, 41 neurons; Csde1 KD2, 23 neurons; Csde1 KD2 + WT, 40 neurons; Csde1 KD2 + Ctnnb1, 28 neurons; Csde1 KD2 + Licl, 23 neurons). Scale bar, 10 μm. (D) Csde1 KD reduced the number of excitatory (vGlut) and inhibitory (vGAT) synapses (NC, 41 neurons; Csde1 KD2, 46 neurons; Csde1 KD2 + WT, 30 neurons; Csde1 KD2 + Ctnnb1, 39 neurons; Csde1 KD2 + Licl, 28 neurons). Scale bar, 10 μm. (E and F) Voltage-clamp whole-cell recordings showed that both frequency and amplitude of mEPSCs and mIPSCs were decreased in Csde1 KD neurons (three neurons for each condition). (G) Immunoblot showed that β-catenin expression was markedly decreased in Csde1 KD neurons. Statistical data were presented as means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

  • Fig. 4 Disruption of dUnr interferes with synapse development and transmission in Drosophila.

    (A) Left: Representative NMJ4 synapses of dUnr-related lines (WT, dunr/+, dunr, dunr/Df1, and dunr/Df2). Right: Both bouton number and satellite bouton number were increased in dunr/+ line (n = 22), dunr line (n = 58), dunr/Df1 line (n = 23), and dunr/Df2 line (n = 25) compared to control (n = 61). The boutons were costained with anti-HRP labeling the neuronal plasma membrane (red) and anti-DLG (green) labeling a postsynaptic scaffold protein. Magnified image of the boxed region at left bottom shows the terminal bouton or the satellite bouton. Scale bar, 5 μm. (B) Left: Representative NMJ4 morphology of the Elav/+ line, a pan-neuronal dUnr KD line (Elav-RNAi), a Drosophila UNR rescue line (dUnr-res), and a human CSDE1 rescue line (hCSDE1-res). Right: Both bouton number and satellite bouton number were increased in Elav-RNAi line (n = 45) compared to Elav/+ line (n = 45). In dUnr-res line (n = 34) and hCSDE1-res line (n = 28), the numbers were decreased compared to Elav-RNAi line (n = 45). Scale bar, 5 μm. (C) Left: Representative traces of EJPs and mEJPs in the indicated genotypes. Right: The amplitudes of EJPs were mildly decreased in both dunr/Df1 line (WT: n = 9; dunr/Df1: n = 13) and pan-neuron KD line (Elav/+: n = 29; Elav-RNAi: n = 17). Slightly but not significantly decreased mEJP amplitude and no significant change of quantal content were observed on both KD and dunr/Df1 lines. NS, not significant. (D) The normalized fluorescent intensity of BRP was slightly increased in dunr (n = 76) and dunr/Df1 (n = 63) lines compared to controls. Scale bar, 5 μm. (E) The normalized fluorescent intensity of postsynaptic GluRIIA was markedly reduced in dunr/Df1 (n = 18), dunr/Df2 (n = 23), and KD lines (n = 25) compared to controls. Scale bar, 5 μm. (F) Decreased normalized fluorescent intensity of FM 1-43 dye was detected in dunr (n = 14), dunr/Df1 (n = 40), and KD (n = 61) lines compared to controls. Scale bar, 5 μm. Statistical data were presented as means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001.

  • Table 1 Summary of CSDE1 LGD variants.

    Isoform, NM_001242891. BCM, Baylor College of Medicine; WES, whole-exome sequencing; gDNA, genomic DNA; cDNA, complementary DNA; SSC, Simons Simplex Collection.

    Patient IDCohortMethodFunctionCohort sizeVariant in gDNA (hg19,chr1)Variant in cDNAProtein changeInheritance
    PU2.p1ProvidenceWESStopgaing.115282407_115282407delGc.243_243delCp.S82Lfs*28Paternal
    BU2.p1BethesdaWESStopgaing.115282401_115282402insTc.248_249insAp.Y83*Paternal
    NN1.p1NijmegenWESFrameshift2,418g.115282363delAc.287_287delTp.F96Sfs*14De novo
    SU2.p1San FranciscoWESStopgaing.115280664G>Ac.367C>Tp.R123*De novo
    SS1.p1SSCWESStopgain2,508g.115280664G>Ac.367C>Tp.R123*De novo
    AA.p1AdelaideTargetStopgain10,745g.115280664G>Ac.367C>Tp.R123*Maternal
    TI.p1TroinaTargetFrameshift10,745g.115275369_115275370TTc.1043_1044delAAp.K348Rfs*12De novo
    BU1.p1BaltimoreWGSStopgain29g.115275305G>Ac.1108C>Tp.R370*De novo
    CC1.p1ACGCTargetStopgain4,045g.115275239C>Ac.1174G>Tp.E392*De novo
    CC4.p1ChangshaTargetSplicing10,745g.115273044C>Tc.1330-1G>APaternal
    PU1.p1BCMWESFrameshift8,910g.115273009_115273010insTc.1363_1364insAp.R455Kfs*3Maternal
    CC2.p1ACGCTargetFrameshift4,045g.115269672_115269673insCc.1533_1534insGV512Gfs*23De novo
    CC3.p1ACGCTargetSplicing4,045g.115269008C>Tc.1603-1G>ADe novo
    TA.p1TASCTargetStopgain10,745g.115268971G>Ac.1639C>Tp.Q547*Paternal
    NN2.p1NijmegenWESFrameshift2,418g.115267916_115267917insCc.1816_1817insGp.D606Gfs*6Not maternal
    SU1.p1BCMWESSplicing8,910g.115267840T>Cc.1891+2A>GPaternal
    TU.p1BCMWESFrameshift8,910g.115261250_115261250delCc.2471_2471delGp.G824Dfs*30Not maternal
    SS2.p1SwedishTargetFrameshift10,745g.115260816_115260819delTCTTc.2506_2509delAAGAp.K836Sfs*17Maternal
  • Table 2 Genotype-phenotype correlations of 17 probands with CSDE1 LGD variants.

    +, present; −, absent; blank, not reported. DN, de novo; MI, maternal inheritance; PI, paternal inheritance; NMI, not maternal inheritance; EEG, electroencephalographic.

    Patient IDPU2.p1BU2.p1NN1.p1SS1.p1SU2.p1AA.p1TI.p1BU1.p1CC1.p1CC4.p1PU1.p1CC2.p1CC3.p1TA.p1SU1.p1TU.p1SS2.p1Total
    Variant inheritancePIPIDNDNDNMIDNDNDNPIMIDNDNPIPINMIMI8 DN,8 INH
    Age at last
    examination (years)
    4.53.512177191113858.433.87.95.25.9103–19
    SexFMMFFMMFMMFMMMMMF11 M, 6 F
    Neurodevelopmental problems
      Developmental
    delay (speech)
    +++++++++++++++++17/17
      Developmental
    delay (motor)
    +++++++++++++++15/17
      ASD/autistic
    features*
    +++±+++++++11/15
      ID++++±++++++±++++±14/16
    Neurological problems
      Epilepsy/seizure±+±±++±7/16
      EEG abnormalities++++++6/12
      MRI brain
    abnormalities
    +++++++7/14
      Macrocephaly++++++6/14
      Sleep
    disturbances
    ++++4/14
    Behavior problems
      Repetitive
    behavior
    ++++++++++++++14/15
      ADHD+++++++++9/13
      Anxiety+++++++7/13
      Obsessive
    behavior
    ++++4/13
      Self-injurious
    behavior
    +++3/12
      Aggressive
    behavior
    ++2/14
    Systemic problems
      Eye
    abnormalities§
    +++++++7/13
      Recurrent
    infections
    ++++++6/13
      Hypotonia++++++6/13
      Hand deformity++++++6/15
      Short stature+++3/16

    *+, ASD; ±, autistic features.

    †+, mild to severe ID; ±, below average or learning disability.

    ‡+, epilepsy; ±, seizure but no epilepsy diagnosis.

    §Eye abnormalities are variable (see table S1).

    ║Hand deformity including brachydactyly (4), polydactyly (1), and clinodactyly (1).

    Supplementary Materials

    • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/9/eaax2166/DC1

      Fig. S1. Mean coverage of the coding regions of CSDE1 in ExAC whole-exome sequencing data.

      Fig. S2. Correlations of the two independent HITS-CLIP experiments and pathway enrichment of Csde1-binding targets.

      Fig. S3. Time-spatial expression pattern of CSDE1 in human and mouse brain.

      Fig. S4. CSDE1 disruptive mutations show loss of function.

      Fig. S5. Interfere efficiency of two shRNA in neurons.

      Fig. S6. Immunoblotting with anti-dUnr antibodies was performed to examine the expression level of dUnr in dunr, dunr/Df1, dunr/Df2, and Da-RNAi lines.

      Fig. S7. Overexpression of dUnr or hCSDE1 has no effect on both bouton number and satellite bouton number compared to WT controls.

      Table S1. Detailed clinical information for probands or carrier patients with LGD mutation or de novo missense mutations.

      Table S2. Validation result of selected RNA binding targets.

      Table S3. High-confidence RNA binding targets called by two software programs in two experiments.

    • Supplementary Materials

      The PDF file includes:

      • Fig. S1. Mean coverage of the coding regions of CSDE1 in ExAC whole-exome sequencing data.
      • Fig. S2. Correlations of the two independent HITS-CLIP experiments and pathway enrichment of Csde1-binding targets.
      • Fig. S3. Time-spatial expression pattern of CSDE1 in human and mouse brain.
      • Fig. S4. CSDE1 disruptive mutations show loss of function.
      • Fig. S5. Interfere efficiency of two shRNA in neurons.
      • Fig. S6. Immunoblotting with anti-dUnr antibodies was performed to examine the expression level of dUnr in dunr, dunr/Df1, dunr/Df2, and Da-RNAi lines.
      • Fig. S7. Overexpression of dUnr or hCSDE1 has no effect on both bouton number and satellite bouton number compared to WT controls.

      Download PDF

      Other Supplementary Material for this manuscript includes the following:

      • Table S1 (Microsoft Excel format). Detailed clinical information for probands or carrier patients with LGD mutation or de novo missense mutations.
      • Table S2 (Microsoft Excel format). Validation result of selected RNA binding targets.
      • Table S3 (Microsoft Excel format). High-confidence RNA binding targets called by two software programs in two experiments.

      Files in this Data Supplement:

    Stay Connected to Science Advances

    Navigate This Article