PQBP1, a factor linked to intellectual disability, affects alternative splicing associated with neurite outgrowth

Abstract

Polyglutamine-binding protein 1 (PQBP1) is a highly conserved protein associated with neurodegenerative disorders. Here, we identify PQBP1 as an alternative messenger RNA (mRNA) splicing (AS) effector capable of influencing splicing of multiple mRNA targets. PQBP1 is associated with many splicing factors, including the key U2 small nuclear ribonucleoprotein (snRNP) component SF3B1 (subunit 1 of the splicing factor 3B [SF3B] protein complex). Loss of functional PQBP1 reduced SF3B1 substrate mRNA association and led to significant changes in AS patterns. Depletion of PQBP1 in primary mouse neurons reduced dendritic outgrowth and altered AS of mRNAs enriched for functions in neuron projection development. Disease-linked PQBP1 mutants were deficient in splicing factor associations and could not complement neurite outgrowth defects. Our results indicate that PQBP1 can affect the AS of multiple mRNAs and indicate specific affected targets whose splice site determination may contribute to the disease phenotype in PQBP1-linked neurological disorders.

Figure 1.

AS targets of PQBP1 in HeLa cells. (A) Schematic of Bcl-x. Light rectangles are exons, with dark color marking the alternatively spliced region, and black lines are introns. Arrows show translation start sites, and arrowheads mark quantitative PCR primer sets. (B) Western blot of HeLa cell lysates where PQBP1 is knocked down by siRNA. The control sample was treated with a siRNA targeting firefly luciferase mRNA. Antibodies are listed on the right. Knockdown (KD) with two different siRNAs is shown. (C) Gel analysis of RT–PCR of Bcl-x splicing isoforms for control and PQBP1 knockdown samples is shown (left panel) with quantification (right panel). Mean of three independent measurements ± SD are shown. Changes were tested by one-way ANOVA. (**) Statistically significant with P-value < 0.01. See also Supplemental Figure S1, A and B. (D) Quantification of real-time PCR data on 14 mRNAs that showed significant AS changes between control and PQBP1 knockdown samples, with Mcl1 (not a target of PQBP1) as the negative control. The Y-axis depicts the ratio between amplicons for constitutive and alternative exonic regions; TPX-2 and CASP6 follow the Y-axis on the right side, while the rest follow the Y-axis on the left side. Light columns show the ratio of two amplicons from the control sample, and dark columns show the ratio of two amplicons from the PQBP1 knockdown sample. Mean of three independent measurements ± SD are shown. The difference of the ratio between control and PQBP1 knockdown samples was analyzed by one-way ANOVA test. (*) Statistically significant with P-value < 0.03; (**) statistically significant with P-value < 0.01; (n.s.) not significant. See also Supplemental Figure S1C.

Figure 2.

The spectrum of proteins associated with PQBP1 and disease-linked mutants. (A) Schematics of the protein domain organization of wild-type (WT) PQBP1 and two disease-linked mutants. Boxes are protein domains, with names shown in the center. (WW) WW domain; (CTD) C-terminal domain. Red arrowheads show the mutation sites in the disease-linked PQBP1 variants. (B) Immunofluorescence staining of HeLa cells stably expressing Flag-HA-tagged wild-type PQBP1, ΔAG, or Y65C. Cells were fixed and stained with antibodies or dyes listed at the bottom of each panel column. (C) Western blot of HeLa cell lines stably expressing empty Flag-HA vector, Flag-HA-tagged wild-type PQBP1, Flag-HA-tagged ΔAG, and Flag-HA-tagged Y65C. The antibodies used are indicated at the bottom of each panel. (D) Silver staining of proteins copurified with Flag-HA-tagged empty plasmid, Flag-HA-tagged wild-type PQBP1, Flag-HA-tagged ΔAG, or Flag-HA-tagged Y65C. (E) GO term enrichments of proteins that were pulled down with wild-type PQBP1. Statistics were calculated with human genes as the background. The X-axis is the corrected P-value (false discovery rate [FDR]) in negative log for enrichment.

Figure 3.

SF3B1 function is affected by PQBP1. (A) Western blot of PQBP1 CLIP together with an irrelevant IgG control, probed with PQBP1 antibody. PQBP1 was not detected in depleted supernatants or the IgG pull-down, indicating efficient and specific antibody binding of PQBP1. (B) Western blot of SF3B1 CLIP from PQBP1 knockdown (KD) samples and samples treated with siRNA targeting firefly luciferase mRNA as control. Shown are total cell lysates of control (lane 1) and PQBP1 knockdown (lane 2) HeLa cells probed with antibodies indicated at the left. (C) Western blot of SF3B1 CLIP from PQBP1 knockdown samples and samples treated with siRNA targeting firefly luciferase mRNA as control. Lanes 1 and 2 are the IgG controls, and lanes 3 and 4 are SF3B1 CLIP stained by SF3B1 antibody for control and PQBP1 knockdown samples. (D) Quantification of Bcl-x and Mcl1 mRNA enrichment in the SF3B1–RNA complex as from SF3B1 CLIP in control and PQBP1 knockdown samples, respectively, by RT–PCR. Data values normalized to the IgG control and mean of three independent measurements ± SD are shown. (E) Proposed model mechanism of PQBP1 AS regulation. A schematic of part of a hypothetical mRNA is shown. Orange rectangles are exons, and the black line is the intron. “GU” marks the 5′ splice site, “A” marks the branch site recognized by the U2 snRNP with help from components like SF3B1, and “AG” marks the 3′ splice site. Green dashed double lines indicate the association between PQBP1 and SF3B1. Through an association with SF3B1 and other splicing factors, PQBP1 influences the recruitment of U2 snRNP to specific target sites, thus affecting splicing decisions on a subset of mRNAs. Loss or mutations of PQBP1 interrupt the splicing factor association and the recognition of specific splice sites. See the Discussion.

Figure 4.

PQBP1 affects dendritic outgrowth and branching in mouse cortical neurons. (A) Western blot of PQBP1 expression in mouse embryonic cortical and hippocampal neurons, compared with endogenous β-actin expression. (B) Immunofluorescence staining of mouse embryonic primary cortical neurons (the four top panels) and rat embryonic primary cortical neurons (the three bottom panels). Neurons were fixed and stained with antibodies or dyes listed at the top of each panel. See also Supplemental Figure S3A. (C) Western blot of mouse embryonic cortical neurons infected with virus from a nontargeting control shRNA or a PQBP1 targeting shRNA. See also Supplemental Figure S3B. (D) Immunofluorescence staining of mouse primary embryonic cortical neurons 5 d after infection with virus from a nontargeting control shRNA or a PQBP1 targeting shRNA. (Green) MAP2 staining; (blue) DAPI staining. See also Supplemental Figure S3C. (E) Comparison of the number of intact nuclei between neuron samples 5 d after infection with virus from a nontargeting control shRNA or a PQBP1 targeting shRNA. Neurons were fixed and stained with DAPI. Healthy and intact nuclei were counted for each sample. Mean of eight independent measurements ± SD is shown. The difference in the number of intact nuclei between control and PQBP1 knockdown samples was analyzed by one-way ANOVA test. (n.s.) Not significant. (F) Comparison of the percentage of Annexin V-stained neurons between samples infected with virus from a nontargeting control shRNA or a PQBP1 targeting shRNA. The number of neurons with Annexin V staining and the percentage of Annexin V-positive neurons were calculated and normalized to neuron samples that were not treated by virus. The Y-axis depicts the fold of the percentage of Annexin V-positive neurons in control or PQBP1 knockdown samples to samples that were untreated. Neurons on day 3 and day 6 after infection/plating were studied. The mean of six independent measurements ± SD is shown. The difference in the percentage of Annexin V-positive neurons between control and PQBP1 knockdown samples was analyzed by one-way ANOVA test. (n.s.) Not significant. (G) Immunofluorescence of individual neurons that were transiently transfected with either a nontargeting control shRNA or a PQBP1 targeting shRNA together with a GFP vector. Neurons were grown on a layer of feeder glial cells and examined 5 d after transfection. (H) Sholl analysis of dendritic branching for neurons transiently transfected with a nontargeting control shRNA or PQBP1 targeting shRNA together with a GFP vector. Concentric circles were drawn at intervals of 5 μm. The number of dendritic branch intersections with each concentric circle was counted (mean ± SE, n = 16–17). The X-axis depicts the distance of the concentric circle from the neuron soma center, and the Y-axis depicts the number of intersections the corresponding circle encounters with dendrites.

Figure 5.

Wild-type (WT) PQBP1 but not the mutants rescues the neuron morphological defects. (A) Schematic of the pLKO.1-derived lentiviral constructs used to generate viruses for the rescue experiments. See also Supplemental Figure S3D. (B) Western blot of neuron samples that were infected with virus from an unmodified PQBP1 targeting shRNA, a nontargeting control shRNA, or PQBP1 targeting shRNA constructs coexpressing exogenous wild-type PQBP1, ΔAG, or Y65C. Antibodies used are listed at the right. Quantification of total PQBP1 level for each sample compared with control is shown below each lane. (C) Immunofluorescence of individual neurons that were transiently transfected with PQBP1 targeting shRNA constructs coexpressing exogenous wild-type PQBP1, ΔAG, or Y65C together with a GFP vector. Neurons were grown on a layer of feeder glial cells. (D–F) Sholl analysis of dendritic branching for neurons that were transiently transfected with a nontargeting control shRNA, an unmodified PQBP1 targeting shRNA, or PQBP1 targeting shRNA constructs coexpressing wild-type PQBP1 (D), ΔAG (E), or Y65C (F) (mean ± SE; n = 10–17).

Figure 6.

PQBP1's AS targets in mouse embryonic cortical neurons. (A) A schematic of a defined AS event and quantification of splicing changes. A hypothetical alternatively spliced mRNA is shown, with light rectangles representing exons and black lines between exons representing introns. The middle exon is the alternatively spliced exon. The 5′ IS is the start of a splicing junction, and the 3′ ES is the end of the junction. The AS event shown here includes two sub-AS junctions that share the same 5′ IS. Light dashed short lines mark RNA-seq reads that are mapped to sub-AS junction 1, and dark solid short lines mark reads mapped to sub-AS junction 2. The distribution of reads mapped to the two sub-AS junctions was used for AS quantification in our analysis. See also Supplemental Figure S4, Supplemental Table S2, and the Supplemental Material. (B) RT–PCR validation of 10 randomly picked AS events identified as PQBP1 targets and two randomly picked AS events identified as non-PQBP1 targets in the computational analysis. Abat1 and Icmt were predicted non-PQBP1 targets. The Y-axis depicts the ratio between isoform 1 and isoform 2. Dlgap4, Tmod3, Adam15, and Dclk1 follow the Y-axis on the right side, while the rest follow the Y-axis on the left side. The mean of three independent measurements ± SD is shown. The differences between control and PQBP1 knockdown (KD) samples were analyzed by one-way ANOVA test. (*) Statistically significant with P-value < 0.05; (**) statistically significant with P-value < 0.01; (n.s.) not significant. (C) GO term functional enrichment of AS targets of PQBP1. Statistics were calculated with all genes that are expressed in mouse embryonic cortical neurons (profiled from RNA-seq data) as the background. The GO term enrichment profile here is specific compared with RNA-seq studies for other proteins in neurons (see the text). The X-axis is the corrected P-value (FDR) in negative log for enrichment. See also Supplemental Table S4.

Figure 7.

PQBP1-modulated AS of NCAM-140 influences neurite outgrowth. (A) A schematic of alternatively spliced region of NCAM-140. Green rectangles are exons, and black lines are introns. The gray rectangle marks the VASE. Arrowheads are primers designed for RT–PCR and electrophoresis. (B) NCAM-140 VASE splicing isoforms for neuron samples infected with virus from a nontargeting control shRNA or a PQBP1 targeting shRNA were analyzed by RT–PCR and gel electrophoresis. See also Supplemental Figure S6. (C) RT–PCR quantification for VASE⁻ and VASE⁺. The ratio of the two isoform levels are compared among neurons infected with virus from a nontargeting control shRNA, an unmodified PQBP1 targeting shRNA, or PQBP1 targeting shRNA constructs coexpressing exogenous wild-type (WT) PQBP1, ΔAG, or Y65C. The mean of three independent measurements ± SD is shown. The differences between samples were analyzed by one-way ANOVA test. (**) Statistically significant with P-value < 0.01. (D) Western blot of neurons infected with virus from a nontargeting shRNA, an unmodified PQBP1 targeting shRNA, a PQBP1 targeting shRNA construct coexpressing exogenous wild-type PQBP1, or a PQBP1 targeting shRNA construct coexpressing exogenous VASE⁻ to restore the ratio of VASE⁻/VASE⁺. (E) RT–PCR quantification for VASE⁻ and VASE⁺ expression levels of neurons infected with virus from a nontargeting shRNA, an unmodified PQBP1 targeting shRNA, or a PQBP1 targeting shRNA construct coexpressing exogenous VASE⁻ to restore the ratio of VASE⁻/VASE⁺. The mean of three independent measurements ± SD is shown. The differences between samples were analyzed by one-way ANOVA test. (**) Statistically significant with P-value < 0.01. (F) Immunofluorescence of individual neurons that were transiently transfected with a PQBP1 targeting shRNA construct coexpressing exogenous VASE⁻ to restore the ratio of VASE⁻/VASE⁺ together with a GFP vector. Neurons were grown on a layer of feeder glial cells. (G) Sholl analysis of dendritic branching for individual neurons that were transiently transfected with a nontargeting control shRNA, an unmodified PQBP1 targeting shRNA, or a PQBP1 targeting shRNA construct coexpressing exogenous VASE⁻ to restore the ratio of VASE⁻/VASE⁺ (mean ± SE, n = 15–17).