Cell-Specific RNA Binding Protein Rbfox2 Regulates Ca_V2.2 mRNA Exon Composition and Ca_V2.2 Current Size

The splicing factor Rbfox2 represses e18a inclusion in a cell line. A, Genome alignments: Rbfox binding motif (U)GCAUG located upstream of e18a (Minovitsky et al., 2005). Putative Rbfox consensus sequence is conserved across all 12 species shown. B, Western blotting shows Rbfox protein in F11 cells transfected with siRNA against Rbfox2 (lanes 1-3) compared to control (lanes 4-6). Anti-Rbfox recognizes RNA binding motif (RRM) of all Rbfox protein isoforms (Gehman et al., 2011). The anti-GAPDH signal from the same membrane stripped and reprobed is shown. The experiment was run in triplicate, lanes 1-3 (siRNA) and lanes 4-6 (control). C, RT-PCR analysis of +18a and Δ18a Ca_V2.2 mRNAs from F11 cells transfected with Rbfox2-specific siRNA. Three independent transfections were analyzed per condition. E18a RT-PCR signal, as a percentage of total, was: 22.5 ± 2.3% (nontransfected F11) compared to 58.4 ± 2.5% in cells transfected with 100 nM siRNA to Rbfox2 (Student’s unpaired t test, p = 0.000129^a). D, RT-PCR analysis of Ca_V2.2 mRNA extracted from nontransfected F11 cells (con), F11 cells transfected with siRNA to cyclophilin B (cycB siRNA), or transfected with nontargeting siRNA. Representative gel is shown with average % e18a signal in PCR amplification of F11 RNA from three experiments. Average +18a signals were similar: 22.5 ± 2.3% (con); 20.3 ± 2.3% (cycB); and 26.0 ± 1.7% (nontargeting siRNA; one-way ANOVA, F = 1.789, p = 0.246^b).

Rbfox2 binding to Ca_V2.2 pre-mRNA is reduced in adult compared to P8-P11 sympathetic ganglia. A, CLIP steps to quantify Rbfox2 binding to its motif in the intron 5’ to e18a in Ca_V2.2 pre-mRNA in vivo. Cell lysis, protein cross-linking, nucleic acid fragmentation, DNase treatment to remove gDNA, Rbfox2 immunoprecipitation with polyclonal RBM9 antibody, RNA isolation from the immunocomplex, and first strand RT followed by qPCR. B, CLIP-RNA samples from SCG of postnatal (P8-P11) and adult mice were used for endpoint RT-PCR. Endpoint PCR products were amplified from first strand cDNA using two sets of primers, one set flanks sequence 5’ to e18a containing the Rbfox2 binding motif (5’ pr-e18a; upper gel) and another other one flanking a sequence 3’ to e18a (3’ pr; lower gel) was used as control. Upper gel, 5’-e18a primers amplify 99-bp products from postnatal and adult samples after DNase treatment both before (input; lane 2) and after Rbfox2-IP (Rbfox2; lane 3). Lower gel, 3’-e18a primers amplify 108-bp products from postnatal (P8-P11) and adult samples after DNase treatment before (input; lane 5) but not after Rbfox2-IP (Rbfox2; lane 6). 5’-e18a and 3’-e18a primers did not amplify products from IgG or H₂O control samples (lanes 4, 7, and 8). Products were separated in 2% agarose gel (one biological triplicate, three mice each condition). Size markers are shown (bp; lanes 1 and 9). C, Melting and standard curves for 5’-e18a and 3’-e18a qPCRs establish the quality and efficiency of the two primer sets used for qPCR data in panel D, Melting curves for 5’-e18a and 3’-e18a PCR products were generated after the qPCR using a temperature gradient from 60°C to 95°C (+0.3°C/step). The single sharp peak corresponds to a single product. Standard curves show Ct values for each qPCR primer set and establish they have the same efficiency (slopes = -3.48 ± 0.07 and -3.51 ± 0.05 for 5’-e18a and 3’-e18a qPCRs, respectively). D, Percentage input method (Haring et al., 2007) used to quantify Rbfox2-IP RNA template by 5’-e18a qPCR and referenced to input before Rbfox2-IP. Each point represents a sample containing six SCG from three mice. The complete experiment (from SCG harvesting to qPCR) was performed in triplicate for each age group (P8-P11 and adult). Mean ± SE in % Rbfox2 relative to input for adult: 0.1872 ± 0.05; P8-P11: 0.5471 ± 0.07, p = 0.02^c. Mean and SE values are indicated by horizontal bars.

Rbfox2 is expressed in sympathetic neurons. A, Representative fluorescence image of dissociated cultured neurons from SCG. Three cells (white arrows) in image shown were injected with Rbfox2 siRNA and dextran fluorescein (green) 24 h before all cells were fixed and exposed to anti-Rbfox2 (magenta). siRNA injected cells show little anti-Rbfox2 signal (white arrows), while control cells (not injected) show strong Rbfox2 expression localized to cell nuclei (scale bars, 20 μm). B, left, Traces of calcium currents induce by voltage step to +20 mV from three different SCG neurons not injected (control, cells 1-3) and three neurons injected with dextran/fluorescein and Rbfox2 siRNA (cells 4-6). Middle, Average current density (filled symbol), median (horizontal line), 25th-75th interquartile range (box), and whiskers (range) are shown for calcium currents in control (not injected) neurons (n = 41) and in neurons injected 24 h prior with Rbfox2 siRNA (n = 22). Mean (median) ± SE in pA/pF were for control: 43.99 (41.6) ± 1.9; and for SCG siRNA injected: 56.61 (52.13) ± 6.75. Right, Cumulative frequency plot for charge density measured from control and Rbfox2 siRNA injected SCG neurons.

Larger current density in cells expressing e18a-Ca_V2.2 compared to Δ18a-Ca_V2.2. A, Peak Ca_V2.2 current−voltage relationships in tsA201 cells expressing Δ18a-Ca_V2.2 or e18a-Ca_V2.2 with Ca_Vβ₃ and Ca_Vα₂δ-₁ subunits. Values are mean ± SE and N values are shown in parentheses. Data were compiled from multiple different transfections and gathered by three different experimenters. Individual current−voltage relationships were fit by Boltzmann-linear functions and Boltzmann activation midpoint (V_1/2) and slope (k) estimated. V_1/2 ± SE (mV); Δ18a-Ca_V2.2: -1.3 ± 0.9; e18a-Ca_V2.2: -4.8 ± 0.6 (Student’s unpaired t test, p = 0.0001^e). k ± SE (mV); Δ18a-Ca_V2.2: 5.2 ± 0.5; e18a-Ca_V2.2 = 4.4 ± 0.25. B, Ca_V2.2 current−voltage relationships in cells expressing Δ18a-Ca_V2.2 or e18a-Ca_V2.2 with Ca_Vβ₂ and Ca_Vα₂δ-₁. Values are average ± SE. C, Cumulative frequency plots of peak Ca_V2.2 current densities measured at 5 mV in cells expressing Δ18a-Ca_V2.2 or e18a-Ca_V2.2 with either Ca_Vβ₂ or Ca_Vβ₃. Bin size was 25 pA/pF for all plots. D, Peak Ca_V2.2 current density values are shown for individual cells (open symbols), mean (filled symbol), median (horizontal line), 25th-75th interquartile range (box), and whiskers (range) in tsA201 cells expressing Δ18a-Ca_V2.2 or e18a-Ca_V2.2 with Ca_Vβ₃ and Ca_Vβ₂. Mean (median) ± SE in pA/pF at 5 mV were for Δ18a-Ca_V2.2/Ca_Vβ₃: 95.9 (66.1) ± 17.1; e18a-Ca_V2.2/Ca_Vβ₃: 194.5 (167.7) ± 19.4; Δ18a-Ca_V2.2/Ca_Vβ₂: 42.9 (32.3) ± 9.4; e18a-Ca_V2.2/Ca_Vβ₂ 53.9 (41.6) ± 7.1 (one-way ANOVA on ranks, Bonferroni corrected p = 0.0001^f). E, Amino acid sequence of wild-type e18a exon (+18a), and two mutants. Seven serines were substituted with alanines (7S/7A) in both mutants; one mutant contains an additional lysine to alanine substitution (7S/7A, K/A). F, Peak Ca_V2.2 I-V relationships in cells expressing Δ18a-Ca_V2.2, e18a-Ca_V2.2 WT, e18a-Ca_V2.2 7S/7A, and e18a-Ca_V2.2 7S/7A, K/A. Mean and SE for each condition and N values in parentheses. Mean (median) ± SE in pA/pF at 5 mV for WTe18a-Ca_V2.2: -248.64 (232.1) ± 24.6; 7S/7A: -203.8 (250.3) ± 25.9; 7S/7A,K/A: -190.2 (170.2) ± 29.9 (one-way ANOVA on ranks, Bonferroni corrected, p_WT:7S/7A = 0.262^g and one-way ANOVA on ranks, Bonferroni corrected, p_{WT: 7S/7A,K/A} = 0.213^g).

Mutation strategy to restrict exon choice at e18a locus of mouse Cacna1b gene. A, Schematic illustrates homologous recombination strategy used to generate two mouse strains that express either +18a-only (Cacna1b^tm3.1Dili) or Δ18a-only (Cacna1b^tm4.1Dili) Ca_V2.2 transcripts. Targeting constructs contained 4.4 kb (left) and 2.3 Kb (right) recombination arms, and a NeoR cassette was used to select for successful recombination events in embryonic stem cells. B, left, Splicing patterns in Cacna1b pre-mRNA of WT, +18a-only, and Δ18a-only mice. Right, RT-PCR products from WT, +18a-only Ca_V2.2, and Δ18a-only Ca_V2.2 mouse brain. Primers were located in e18 and e19 as described in Figure 1. Expected bands were observed: two different size PCR products corresponding to +18a and Δ18a from WT, one band corresponding to +18a from +18a-only samples, and one band corresponding to Δ18a from Δ18a-only samples.

Ca_V2.2 currents in +18a neurons are larger compared to Δ18a-only neurons. A, left, Cumulative frequency plots of peak Ca_V current density recorded at 20 mV in SCG neurons from Δ18a-only and +18a-only mice. Right, peak Ca_V current density at 20 mV (pA/pF). Mean (median) values ± SE were for Δ18a-only: 61.54 (60.1) ± 3.24; +18a-only: 73.3 (78.5) ± 6.94 (Mann-Whitney rank sum test p = 0.035i^h). B, left, Ca_V currents evoked by voltage steps to +20 mV from -80 mV in the absence (Con) and presence of 2 μM ω-conotoxin GVIA (Ctx) in acutely dissociated SCG neurons. C, Ca_V2.2 (Ctx-sensitive; left) and non-Ca_V2.2 (Ctx-resistant; right) currents in SCG neurons from Δ18a-only (black) and +18a-only (red) mice. Values for Ca_V2.2 current density mean (median) ± SE for Δ18a-only: -39.43 (-37.8) ± 4.8 pA/pF; +18a-only: 63.1 (-64.0) ± 5.2 (Student’s t test, unpaired p = 0.009ⁱ). Right, Values for non-Ca_V2.2 current density mean (median) ± SE for Δ18a-only: -31.1 (-31.0) ± 2.3 pA/pF; e18a-only: -27.5 (-26.2) ± 2.2 (Student’s t test, unpaired p = 0.353ⁱ). Experimenter was blind to genotype until after all recordings and analyses were complete. All box plots: values shown are for individual cells (open symbols), mean (filled symbol), median (horizontal line), 25th-75th interquartile range (box), and whiskers (range).