Cacophony alternative splicing gives rise to different protein isoforms.

(A) Cacophony is located on the non-coding strand of the X-chromosome between positions 11,931,280 and 11,985,087 (Flybase version r6.59). Reading direction is indicated on the top right (arrow). Horizontal black lines indicate introns, larger introns are broken by dots. Black boxes indicate exons shared by all cacophony splice variants, red boxes indicate alternative exons, green boxes represent exons that are mutually exclusively spliced. Enlarged area on the right emphasizes mutually exclusive exon pairs IS4A/B and I-IIA/B. (B) Schematic of cacophony with N-terminal sfGFP or mEOS4b tag and two exon pairs that are spliced mutually exclusively. IS4A and IS4B exons encode isoforms of the 4th transmembrane domain (S4) and thus part of the voltage sensor of the first homologous repeat (I) of the calcium channel, while I-IIA and I-IIB give rise to two versions of the intracellular linker between homologous repeats I and II. I-IIA contains a non-well conserved binding site for voltage gated calcium channel β-subunits (Caβ) whereas I-IIB gives rise to a conserved Caβ-binding site as well as a binding site for G-protein βγ-subunits (Gβγ). (B) Genomic removal of one of the mutually exclusively spliced exons of one or two exon pairs by CRISPR/Cas9 mediated double strand breaks in the germ line results in exon out mutants by imprecise excision of the cut exons and cell-intrinsic DNA repair. (C) Cacophony gives rise to 18 annotated transcripts (RA-RU, left). Multiple variants express the same mutually exclusive exons but differ with respect to expression of other alternatively spliced but not mutually exclusive exons. Removal of the mutually exclusive exons IS4A/IS4B and/or I-IIA/I-IIB allows expression of fewer cacophony splice variants. Transcript variants that are possible upon exon excision are marked by +. (D) Western Blots reveal expression of GFP-tagged cacophony protein (cacsfGFP) for all excision variants at the expected band size of ∼240 kDa (cacophony) plus ∼30 kDa (sfGFP, top), while no band can be detected in the Canton S wildtype (CS, top, right) that does not express cacsfGFP. 10 adult brains were used of hemizygous males for all genotypes except for ΔIS4B which is homo-/hemizygous lethal. For ΔIS4B 20 brains of heterozygous females and a heterozygous cacsfGFP control were used (F1 females from cross with Canton S wildtype flies(+)). β-actin was used as loading control (bottom). ΔIS4B shows weak expression (top, left), while all other exon out variants express strongly, although ΔI-IIB shows somewhat weaker expression (top, middle).

The IS4B exon is required for cacophony localization to AZs and for evoked synaptic transmission.

(A-C) Representative confocal projection views of triple labels for GFP tagged cac channels (green), the AZ marker brp (magenta), and HRP to label axonal membrane (blue) as well as enlargements of the area marked by dotted rectangles (Ai-Ci). Label was done in control animals with all cac exons (cacsfGFP, top row, A, Ai), in animals with selective excision of either the alternative exon IS4A (ΔIS4AsfGFP, middle rows, B, Bi), or the alternative exon IS4B (ΔIS4BsfGFP, bottom row, C, Ci). Excision of IS4B is embryonic lethal, so that localization analysis was conducted in heterozygous animals (ΔIS4BsfGFP/+). The gross morphology of the neuromuscular junctions (muscle fibers, bouton numbers and sizes, AZ numbers) was similar in all three genotypes. GFP tagged cac channels localize to AZs (A, Ai) as previously reported (Gratz et al. 2019; Krick et al. 2021). Excision of the IS4A exon does neither impact cac AZ localization nor labeling intensity (B, Bi). By contrast, upon excision of the IS4B exon, no cac label is detected (C, Ci). (D) Pearson’s correlation analysis of cac and brp in cacsfGFP (green), ΔIS4A (red), and heterozygous ΔIS4B (dark green) animals reveals correlation coefficients of ∼0.6 for cacsfGFP and ΔIS4A but not for ΔIS4B (Pearson’s correlation coefficient ∼0.26, which is significantly different from cacsfGFP control (p<0.001, ANOVA with Dunnett’s post hoc test, comparison only against control) and considered negligible. This is in line with absent label in ΔIS4B animals (C, Ci). (E, F) Representative traces of evoked synaptic transmission as recorded in TEVC from muscle fiber 6 upon extracellular stimulation of the motor nerve from a wildtype control animal (CS, blue), an animal cacsfGFP (green), and animals with cacsfGFP and either IS4A exon excision (ΔISA4, red) or transheterozygous female animals with IS4B excision over IS4A excision (ΔISAB/ΔISA4, black). (E) Excitatory postsynaptic currents (EPSCs) are similarly shaped between CS control (blue) and animals expressing cacsfGFP (green), and (F) EPSC amplitudes are not statistically different (p=0.34, two sided Tukey’s multiple comparison test). In animals with homozygous IS4A exon excision (red), EPSC amplitude is slightly but not significantly increased (p=0.18, two sided Tukey’s multiple comparison test). In transheterozygous animals with IS4A excision on one chromosome and IS4B excision on the other one, EPSC amplitude is significantly decreased (p=0.0008), two sided Tukey’s multiple comparison test). (G) Quantal size (mEPSC amplitude) and spontaneous release frequency (H) show no significant difference between genotypes.

The IS4B exon is required for cacophony localization to AZs and for evoked synaptic transmission

(A, B) Since animals homozygous for IS4B exon excision are lethal, we created mosaic animals that were heterozygous for cac in most neurons but hemizygous for either cacsfGFP or ΔIS4B in motoneurons innervating muscle M12 (see methods, cacFlpStop). In controls with all cac exons (cacsfGFP, green, top row), cacsfGFP colocalizes with brp (magenta) in presynaptic AZs on M12 (A, enlargements of area indicated by dotted white rectangle in bottom right corner of each image) and evoked synaptic transmission induces EPSCs of about 100 nA amplitude (B). By contrast, upon deletion of IS4B (A, bottom row, see also enlargement in bottom right corner) in motoneurons to M12 no cac label is found throughout the motor terminals that are marked by brp (magenta) and evoked synaptic transmission is reduced by more than 90%, Student’s T-test, p < 0.0001 (B). (C) Cacophony mediates HVA as well as LVA calcium currents in adult flight motoneurons. All currents are recorded from the somata of adult flight motoneurons in mosaic animals with only one copy of the cac locus in flight motoneurons (see methods). HVA currents (upper traces) are measured by starting from a holding potential of –50 mV (LVA inactivation) followed by step command voltages from –90 mV to +20 mV in 10 mV increments (left, top row), while total cac current (lower traces) is elicited by step commands in 10 mV increments from a holding potential of –90 mV allowing activation of LVA currents (left, bottom row). In GFP-tagged controls (cacsfGFP / cacFlpStop) this reveals transient and sustained HVA current components (middle, top traces). However, following excision of the IS4B exon (ΔISABsfGFP / cacFlpStop), the sustained HVA current is absent. By contrast, upon excision of IS4B, the total cac current that also contains cac LVA currents is only partially decreased (middle, bottom traces). Current-voltage (IV) relation of sustained HVA and for total cac current for controls with all cac exons (cacsfGFP, green circles, n = 7) and following excision of IS4B (ΔIS4BsfGFP, dark green squares, n = 4).

Excisions at the I-II exon do not affect AZ cacophony localization but can alter cacsfGFP label intensity in AZs and EPSC amplitude.

(A-C) Representative confocal projection views of triple labels for GFP tagged cac channels (green), the AZ marker brp (magenta), and HRP to label axonal membrane (blue) with enlargements that are indicated by white dotted rectangles in A-C (Ai-Ci). Label in control animals with all cac exons (cacsfGFP, top two rows, A, Ai), with selective excision of either the alternative exon I-IIA (ΔI-IIAsfGFP, middle two rows, B, Bi), or the alternative exon I-IIB (ΔI-IIBsfGFP, bottom two rows, C, Ci). The gross morphology of the neuromuscular junctions (muscle fibers, bouton numbers and sizes, AZ numbers) was similar in all three genotypes (not shown). Excision of the I-IIA exon does neither impact cac AZ localization nor labeling intensity (B, Bi, H). Excision of I-IIB does not impact cac AZ localization but labeling intensity seems lower (C, Ci, H). (D-F) Quantification of cac co-localization with the AZ marker brp yields a similar Pearson’s colocalization coefficient (D) as well as similar Manders 1 (E) and Manders 2 (F) coefficients for controls (green) and both exon-out variants of the I-II locus (ΔI-IIA purple, ΔI-IIB orange) as well as for ΔIS4A (red) but not for ΔIS4B (dark green, ANOVA with Dunnett’s post hoc test, p < 0.0001). (G) ΔI-IIBsfGFP shows fainter immunofluorescence signals in the AZ as compared to control (cacsfGFP) and ΔI-IIAsfGFP. (H) Quantification confirms a significant reduction in ΔI-IIBsfGFP labeling intensity (Kruskal Wallis ANOVA with Dunn’s post hoc test, p < 0.0001) and no differences between ΔI-IIAsfGFP and control (p > 0.99). (I) Evoked synaptic transmission as recorded in TEVC from muscle fiber 6 upon extracellular stimulation of the motor nerve. Excitatory postsynaptic currents (EPSCs) are of similar shape and amplitude for CS control (blue) and animals with GFP-tagged cac channels (cacsfGFP, green, p=0.34, two sided Tukey’s multiple comparison test). Excision of I-IIA (ΔI-IIAsfGFP, purple) has no effect on evoked release amplitude (p=0.52, two sided Tukey’s multiple comparison test), but excision of I-IIB (ΔI-IIBsfGFP, orange) reduces evoked release significantly (p < 0.0001). (J) Quantification of EPSC amplitude reveals a highly significant reduction in ΔI-IIBsfGFP (orange) as compared cacsfGFP controls (green), but neither animals with excision of the I-IIA exon (purple), nor transheterozygous animals with excision of I-IIA on one and I-IIB on the chromosome (brown) show differences to control (p=0.97, two sided Tukey’s multiple comparison test). (K) Quantal size (mEPSC amplitude) and spontaneous release frequency (L) show no significant difference among genotypes.

Dual color STED imaging reveals equal nanoscale channel localization in AZs of cacophony for all exon-out variants, and live sptPALM imaging reduced channel numbers in AZs for ΔI-IIB.

(A) Representative intensity projection image of the AZ marker bruchpilot (labeled with anti-brp, green) and cac clusters (cacsfGFP labeled with anti-GFP, magenta) as imaged with dual color STED at motoneuron axon terminal boutons on larval muscle M6. The dotted white box demarks one bouton that is enlarged in (B). Each cac cluster (magenta) is in close spatial proximity to the AZ marker brp (green) and depending on AZ orientation, cac and brp is viewed from different angles. Top views (= planar views, see C1 in B and in selective enlargement) show 4 brp puncta that symmetrically surround the central cac cluster. Viewing AZs at the edge of the bouton shows the cac cluster facing to the outside and the brp puncta in close proximity (see 2-6). (C) Selective enlargements of each AZ that is numbered in B. (D) Top views (left column) and side views (right column) of the cac/brp arrangement in AZs in controls with cacsfGFP (top row), with excision of exon IS4A (ΔISA4sfGFP, second row), with excision of exon I/IIA (ΔI/IIAsfGFP, third row), and with excision of exon I/IIB (ΔI/IIBsfGFP, bottom row). (E) Quantification of the distances between the center of each cac punctum to the nearest brp punctum in the same focal plane. (F-G) Live sptPALM imaging of mEOS4b tagged cac channels from AZs of MN terminals on muscle 6 in controls with full isoform diversity (cacmEOS4b, green) and following the removal of either I-IIA (ΔI-IIAmEOS4b, purple) or I-IIB (ΔI-IIBmEOS4b, orange). (F) Quantification of channel numbers from bleaching curves (G) reveals ∼ 9-11 cacophony channels per AZ for tagged controls, which matches previous reports (Ghelani et al. 2023). Counts for ΔI-IIA reveal no significant differences (Kruskal-Wallis test with Dunn’s posthoc comparison, p=0.94), but cac number in AZs is reduced by ∼50 % in ΔI-IIB (p<0.0001). (G) Bleaching curves of single AZs were illuminated after ∼10 s and then imaged under constant illumination for another 240 s. Discrete bleaching steps (dotted lines) indicate the bleaching of single cacmEOS4b channels. Comparing the amplitudes of single events and their integer multiples (dotted lines) to the maximum fluorescence at illumination start allows estimates of the total channel number per AZ.

Alternative splicing in the I-II linker affects short term plasticity due to decreased calcium influx and motor behavior.

(A-D) Paired pulse ratio (PPR, ratio of second EPSC amplitude divided by first EPSC amplitude) as measured in 0.5 mM external calcium (A-C) or in 1.8 mM external calcium (D) at different interpulse intervals (IPIs ranging from 10 ms to 100 ms) in control animals (cacsfGFP, A), in animals with removal of I-IIA (ΔI-IIAsfGFP, B), and in animals with removal of I-IIB (ΔI-IIBsfGFP) either in 0.5 mM calcium (C) or 1.8. mM calcium (D). The large variance in PPR upon excision of I-IIB (C) is rendered control-like if the first EPSC amplitude of the twin pulse is adjusted to 0.5 mM external calcium control level (D; comp. with A). This is also reflected in the coefficient of variation (COV) for PPRs (E, 0.5 mM calcium: cacsfGFP green, ΔI-IIA purple, ΔI-IIB orange; 1.8 mM calcium: ΔI-IIB orange/black pattern). (F-I) Synaptic depression as measured in 0.5 mM external calcium in response to stimulus trains of 1 minute duration at 1 Hz frequency for animals with GFP-tagged cac (cacsfGFP, F), following removal of I-IIA (ΔI-IIAsfGFP, G), and with removal of I-IIB (ΔI-IIBsfGFP, H). The top traces show representative TEVC recordings from the postsynaptic muscle cell, and the diagrams mean values (n=5 for F and G, N=6 for H, error bars are SD). (H) The light gray trace shows ΔI-IIB in 1.8 mM external calcium, while the black trace shows ΔI-IIB in 0.5 mM calcium. For all 3 genotypes, depression reaches steady state at ∼ 80 % of the original EPSC amplitude, but upon excision of I-IIB it is more variable (H). Depression time courses do not differ between genotypes but are more variable in ΔI-IIB, independent of external calcium concentration (H, I). (J-M) Synaptic depression in response to stimulus trains at 10 Hz frequency for animals with GFP-tagged cac (cacsfGFP, J), following removal of I-IIA (ΔI-IIAsfGFP, K), and with removal of I-IIB (ΔI-IIBsfGFP, L: black in 0.5 mM calcium, gray trace in 1.8 mM calcium). Again, depression is most variable between animals upon excision of I-IIB (L, M) but time courses do not differ between genotypes. However, time course variation decreases in 1.8 mM calcium in animals with excision of I-IIB (M). Motoneuron stimulation at 60 (N) or 120 Hz (O) frequency, both for durations of 200 ms in animals with GFP-tagged cac (cacsfGFP, top traces), following removal of I-IIA (ΔI-IIAsfGFP, middle traces), and with removal of I-IIB (ΔI-IIBsfGFP, bottom traces). To compare charge transfer across the NMJ during high frequency bursts the total EPSC area below baseline (prior to stimulation) was measured during each 200 ms burst and plotted for each genotype for 60 Hz stimulation in (P) and for 120 Hz stimulation in (Q). Decreased charge transfer in animals with excision of the I-IIB exon is rescued to control level if external calcium is increased to 1.8 mM so that the first EPSP matches control amplitude in 0.5 mM external calcium (P, Q, far right data points ΔI-IIB in 1.8 m calcium). (R) shows single evoked EPSC half amplitude width. (S-V) show different measurements during larval crawling for control animals with GFP-tagged cac (cacsfGFP), removal of I-IIA (ΔI-IIAsfGFP), removal of I-IIB (ΔI-IIBsfGFP), and in transheterozygous animals with removal of I-IIA on one and removal of I-IIB on the other chromosome (ΔI-IIAsfGFP/ΔI-IIBsfGFP). The measured parameters are mean speed during 10 minutes of crawling (S), mean speed without any stops (T), the relative time spent stopping (U) and the maximum speed reached (V). In all diagrams each dot demarks a measurement from a different animal and horizontal bars the medians. For statistics, non-parametric Kruskal Wallis ANOVA with planned Dunn’s posthoc comparison to control was conducted.

Removal of I-IIB impairs presynaptic homeostatic potentiation.

(A-C) Acute presynaptic homeostatic potentiation (PHP) can be induced in control (cacsfGFP) and ΔI-IIA but not in ΔI-IIB animals by bath application of the glutamate IIA receptor (GluRIIA) blocker philanthotoxin (PhTx) (A-C; black traces: without PhTx, red traces after PhTx application). (D-F) Quantification of EPSC amplitude, mEPSC amplitude, and mean quantal content (mQC) is shown as % change in PhTx treated animals to untreated control within genotypes (% Baseline (−PhTx). Untreated controls are set to 100 % (dotted line in D-F). EPSC amplitudes in cacsfGFP and ΔI-IIA animals are not significantly affected (D, cacsfGFP, green p=0.63, ΔI-IIA, purple p=0.44), while the EPSC amplitude in ΔI-IIB animals is significantly reduced after PhTx treatment (D, orange, p=0.0014). However, reduction of mEPSC amplitude in all genotypes shows successful block of GluRIIA compared to the respective control (E, % change to untreated control within genotype; cacsfGFP, green p=0.012; ΔI-IIA, purple p=0.0015, ΔI-IIB, orange p=0.0012). Accordingly, mean quantal content (mQC) is increased in cacsfGFP and ΔI-IIA animals but not in ΔI-IIB animals (F, cacsfGFP, green p=0.084; ΔI-IIA, purple p=0.0004, ΔI-IIB p=0.76). (G) PHP maintenance is typically assessed in GluRIIA mutants. PHP maintenance is observed in cacsfGFP animals because mQC is increased in a GluRIIASP16 mutant background (GluRIIASP16; green, p<0.0001). By contrast, in ΔI-IIB animals the GluRIIA mutant background does not cause an increase in mQC (orange p=0.17). All pairwise comparisons in (D-G) use unpaired Student’s T-test between untreated and treated condition within each genotype. (H) In animals that are transheterozygous for the removal of I-IIA and the removal of I-IIB, and carry GFP-tagged I-IIA and mEOS4b-tagged I-IIB cacophony channels (I-IIAsfGFP/I-IIBmEOS), triple immunolabel for the AZ marker brp, GFP, and mEOS4b show that most AZs (blue) but not all (white arrow heads in H1, H2 and asterisk in H2) contain both, GFP-tagged I-IIA (green) and mEOS4b-tagged I-IIB (magenta) channels. Magenta label in the overlay (H1, H2, right column, arrow heads) indicates AZs with only I-IIB channels, green label in the overlay (H2, right column, asterisk) indicates one AZ with only I-IIA channels. (I) Quantification shows that >95 % of all brp positive AZs contain I-IIA and I-IIB channels (gray dots), few AZs (∼5 %) contain only I-IIB (= ΔI-IIA; purple), and almost no AZ contains only I-IIA (= ΔI-IIB; orange). (J) Relative fluorescence intensity of sfGFP-tagged I-IIB channels (= ΔI-IIA; purple) and sfGFP-tagged I-IIA channels (= ΔI-IIB; orange). For assessment of I-IIB channels, tagged channels were expressed transheterozygously over untagged I-IIA channels (ΔI-IIAsfGFP/ΔI-IIBno tag; purple) and vice versa (ΔI-IIBsfGFP/ΔI-IIAno tag; orange). Quantification of relative fluorescence intensity compared to transheterozygous control (cacsfGFP/cacno tag) reveals an expression ratio of I-IIB to I-IIA channels of 2:1.

gRNA sequences used for cas9 target site.

Vertical lines depict intended break points. Bold and underlined nucleotides indicated PAM (protospacer adjacent motif) sequences.

Cycler settings for exon out verification.

Exon out verification primers

CacophonyIS4A channels are sparsely expressed in the larval brain and ventral nerve cord.

(A, B) Antibody labeling of GFP-tagged cacophony channels lacking exon IS4B (ΔIS4BsfGFP) reveals sparse cac label in the larval brain (green, A-Aiv, see arrows in Aiii) and the larval ventral nerve cord (green, B-Bii, see asterisks and arrow heads in B). Label was conducted in transheterozygous animals expressing cacsfGFP IS4A over untagged cacIS4B (ΔIS4BsfGFP/ΔIS4Ano tag) to avoid expression of untagged IS4A channels. This likely increases labeling intensity. Cac label co-incides with AZ label as indicated by brp label (magenta, Ai, Aiv; Bi, Bii). (C, D) To exclude the possibility that cac label is misinterpreted due to overlap of with brp channel, antibody label was repeated in untagged Canton S wildtype animals (C-CiV, D-Dii). Even strong brp label does not show in the green channel (C, Ciii, D), thus indicating that cacIS4A channels are indeed made and expressed. Images are maximum projections off F 13 single images, (B) 7 single images, (C) 10 single images, (D) 8 single images, z-size is 1 µm.

GluRIIA intensity across the NMJ is not altered upon excision of I-IIA or I-IIB.

(A-C) GluRIIA is expressed in all NMJs independent of cacophony I-II exon excision (green; A, Aii: control, B, Bii: ΔI-IIA, C, Cii: ΔI-IIB, see overlap of GluRIIA and brp in Aii-Cii). The AZ scaffold protein brp (Ai-Ci) was used as mask to assess GluRIIA labeling intensity (D). Quantification of GluRIIA labeling intensity is shown as mean gray value and shows no difference between genotypes (D, cacsfGFP control green, ΔI-IIAsfGFP purple, ΔI-IIBsfGFP orange, ANOVA with Dunn’s pairwise comparison).