Assay for Activity-Induced Alternative Splicing in Primary Neurons and Glia Using Reporters Driven by Cell-Specific Promoters
Primary neurons, due to their responsiveness and plasticity, should represent a useful model system to study activity-induced changes in alternative splicing, but to date, such systems have not been well developed. In this study, we have developed low-density cortical cultures from the embryonic rat forebrain as an experimental system to address the molecular mechanisms of induced changes in alternative splicing using cell-specific and biochemical approaches. Upon dissociation and plating in culture, the postmitotic neurons extend processes, establish synaptic connections, and become electrically active, whereas glial cells in these cultures play supportive roles.
Initially, changes in endogenous splicing patterns were monitored in the cortical cultures both as a function of KCl-induced depolarization, and cell differentiation using the CI cassette exon as a readout. The cortical cultures at 6, 8, 12, and 14 d in vitro (DIV) after plating were treated for 24 h with media containing 25 or 50 mM KCl, since these conditions are known to induce membrane depolarization in these cultures. Changes in the percent exon inclusion values were calculated as the difference between the value in the 50 mM KCl sample and the mock-treated control (ΔEI values).
Each stage of differentiation of the cultures showed similar trends in which CI cassette exon inclusion decreased (ΔEI, −32% to −36%; , lanes 1–12) as a function of the KCl treatment. Based on the consistent response of each exon in this time frame, we used the cultures to ask whether KCl treatment produces transient or stable changes in the splicing patterns of the endogenous RNA transcripts. If transient changes in transcriptional or post-transcriptional events affect splicing factors involved in these mechanisms, we would expect to see a reversal to the basal splicing pattern with time after removal of the stimulus. Reversibility was tested in the 12-d cultures, again using the CI cassette splicing events as a readout. After treatment with KCl, the cultures were changed to fresh medium without KCl, and cells were harvested 0, 6, 12, and 24 h later. KCl-induced changes in the splicing of the CI cassette exon progressively lessened with time to approximately basal levels after KCl washout (ΔEI, −8%; , lanes 13–24).
Reversibility and Effects of Neuronal Differentiation in the Response of the CI Cassette Exon to KCl-Induced Depolarization in Rat Cortical Cultures
For comparison, we assayed for the effects on alternative exon 9 (E9) of the GABAA
receptor γ2 (GABARG2
gene) in the same cultures, since E9 is known to undergo silencing by the polypyrimidine tract binding protein, but not hnRNP A1 (Figure S1
A). Interestingly, E9 inclusion was responsive to the KCl treatment, but this effect was maximized at longer differentiation times of the cultures, and, in contrast to the CI cassette, E9 inclusion increased (ΔEI, 23%, lanes 1–12). These effects were also largely reversible after KCl washout (lanes 13–24). Thus, the KCl-induced effects on CI cassette and E9 inclusion are largely reversible, and suggest a role for mRNA homoeostasis in allowing for the readjustment of splicing patterns to basal levels.
To experimentally identify nucleotide sequence requirements for activity-induced alternative splicing in neuronal and glial cell types in the cortical cultures, we turned to cell-specific promoters that were previously characterized in transgenic mice. For this purpose, we assessed the ability of the alpha CaMKII promoter to drive neuron-specific expression in rat cortical cultures, since this promoter was shown previously to drive forebrain-specific expression in excitatory neurons of transgenic mice [26
]. Fluorescent reporters were constructed in which the expression of enhanced yellow fluorescent protein (EYFP) was driven by various portions of the CaMKII promoter (, constructs 1 and 2). Construct CaMKII_279 EYFP contained an 8.5-kilobase (kb) promoter region known to restrict expression of Cre to the forebrain of transgenic mice. For cloning purposes, a shorter promoter region was also tested, which contained the start site–proximal 2.2-kb promoter fragment (CaMKII_22 EYFP).
Promoters from Transgenic Mice Drive Neuron- and Glial-Specific Expression in Cortical Cultures
To determine expression patterns in the cortical cultures, confocal microscopy was used to measure the extent of overlap between EYFP expression and antibody staining with TRITC-labeled NeuN or glial fibrillary associated protein (GFAP), which served as molecular markers for neurons and glia, respectively (B). For comparison, we also constructed a fluorescent reporter in which DsRed was fused to a 2-kb region of the Gfa2 promoter, Gfa2 DsRed (, construct 3) [27
]. The Gfa2 promoter was shown previously to drive expression predominantly in glial cells of transgenic mice [28
]. As controls, the EYFP and DsRed coding sequences were also fused to the human cytomegalovirus immediate early (CMV) promoter (, constructs 4 and 5).
The strong neuron specificity of expression of the CaMKII promoters in these cultures is shown as the percent overlap expression of each construct with NeuN or GFAP antibody staining (A, top right). For each sample, a minimum of 200 cells with EYFP fluorescence were examined for TRITC-labeled NeuN staining. Whereas, the largest (8.5 kb) region of the promoter (CaMKII_279) showed 96% overlap with NeuN and 3% overlap with GFAP, similar results were observed with the 2.2-kb region (CaMKII_22; 95% overlap with NeuN and ~5% overlap with GFAP). In contrast, the Gfa2 promoter was preferentially expressed in glia in these cultures (94%), and some overlap with neurons was observed (22%) in agreement with previous analysis in transgenic mice [28
]. A summary of these results is also shown graphically (C). As a reference, EYFP expressed from the CMV promoter (CMV_EYFP) showed 64% overlap with NeuN and 21% overlap with GFAP, whereas the DsRed expressed from the same promoter (CMV_DsRed) showed 54% overlap with NeuN and 26% overlap with GFAP. Due to its mixed expression profile, the CMV promoter was not used for further analysis.
Representative examples of the overlap expression of the two CaMKII EYFP reporters with NeuN is shown in B. Overlap expression is indicated by yellow fluorescence in the nuclei of samples stained with TRITC-labeled anti-NeuN (CaMKII EYFP reporters, anti-NeuN panels), and the lack of overlap expression in samples containing the anti-GFAP antibody is clearly shown (anti-GFAP panels). The pattern of expression for samples containing the Gfa2 DsRed reporter characteristically showed overlap of DsRed fluorescence with FITC-labeled anti-GFAP, but not with FITC-labeled anti-NeuN (Gfa2 DsRed panels).
To confirm the neuron- versus glial-specific expression patterns of the promoters, we co-transfected the CaMKII_22_EYFP and Gfa2-DsRed plasmids into the same cortical cultures and determined overlap expression. No greater than 22% overlap would be expected based on the least specific of the two promoters, Gfa2. The observed overlap ranged from 14% to 19% (raw values), in contrast to 82% to 84% from the generally expressed CMV promoters. After normalizing these values to 100% transfection efficiency for the CMV samples, the co-expression of CaMKII and Gfa2 promoters ranged from 17% to 23% (corrected values). These results are in good agreement with the promoter selectivity shown above.
We next asked how these promoters affect splicing patterns in established neuronal (PC12 and ST15A) and muscle myoblast (C2C12) cell lines, in comparison to the primary cortical cultures. Similar CI cassette exon inclusion levels were observed from the two promoter types in the neuronal and non-neuronal cell types (ΔEI, <3.6%, ). Thus, the increase in CI cassette exon inclusion arising from the promoter fusions in primary cortical cultures must reflect the properties of the neurons and glial cells in the cultures, and cannot be explained by the promoters alone.
Effects of Cell-Specific Promoters on Alternative Exon Inclusion in Cell Lines versus Cortical Cultures for the Splicing Reporter CI wt0
Next we took advantage of the CaMKII and Gfa2 splicing reporters to characterize the KCl-induced effects on CI cassette exon silencing as a function of cell type in the primary cortical cultures. Wild-type CI cassette splicing reporters driven by the cell-specific promoters CaMKII_22 CI wt0 and Gfa2 C1 wt0 were transiently expressed in cortical cultures for 18 h, followed by treatment with KCl for 6 and 24 h. When transcripts were expressed from the CaMKII_22 promoter, CI cassette exon inclusion decreased in response to the KCl treatment, and this effect was consistent in neurons at all time points examined (A, lanes 7–12). Similar effects were observed for the endogenous CI cassette exon when mock-transfected cultures were treated in parallel (lanes 1–6). Optimal effects were reached after 24 h for both endogenous and CaMKII_22-expressed transcripts. In contrast, when transcripts were expressed from the Gfa2 promoter, CI inclusion showed little or no change at the 6- and 24-h time points (lanes 13–18). Based on the cell specificity of the CaMKII promoter shown above, we conclude that neurons play the predominant role in the KCl-induced changes in alternative splicing in these cultures. Furthermore, the trend and magnitude of the effects for the exogenous CI cassette exon inclusion are in good agreement with those of the endogenous transcripts. As a corollary to this experiment, we measured the number of CaMKII_EYFP-expressing neurons that fail to exclude trypan blue. In mock- and depolarization-treated cultures, we observed that the viability of the neurons was high in both the mock-treated (94% viable) and depolarization-treated (93% viable) samples (B). Thus, we conclude that the RT-PCR results shown above (A, lanes 7–12) largely reflect splicing pattern changes in viable neurons.
Cell-Specific Splicing Reporter Assay for Depolarization Induced Alternative Splicing and Effects on Cortical Neuron Viability
To further validate these results, we transferred the E9 splicing reporter into the CaMKII and Gfa2 promoter constructs and tested their response to KCl treatment in the cortical cultures (Figure S1
B). When expression was driven in neurons, E9 inclusion increased similarly to that of the endogenous transcripts (ΔEI, 17% and 15%, respectively), but this was not the case when expression was driven in glial cells (ΔEI, 3%). Thus, for a second test exon, the promoter system recapitulated the splicing response in neurons as observed for the endogenous transcripts.
Identification of RNA Sequence Motifs That Mediate the Response of Exon Skipping to Neuronal Activity
Based on the well-behaved exon-skipping response of the CI cassette exon to KCl-induced depolarization in this system, we focused on this exon to identify nucleotide sequences involved in the response. A three-component code of two exonic UAGGs and an intronic GGGG motif, which was recently shown to mediate CI cassette exon silencing [29
], served as the starting point for these experiments. Splicing reporters with point mutations in combinations of the three silencing motifs were subcloned under the control of the CaMKII_22 promoter, and assayed in the cortical cultures (A).
Nucleotide Sequence Requirements for KCl-Induced Depolarization Effects on Splicing in Primary Neurons
In the absence of KCl treatment, the CI wt0 reporter with intact UAGG and GGGG motifs showed strong silencing activity (exon skipping) in the primary neurons in agreement with our previous analysis in cultured cell lines. Compared to the wild-type substrate, point mutations in a single silencer increased exon inclusion (A, lanes 1, 4, 7, and 13), and multiple point mutations generated complete or nearly complete exon inclusion (lanes 10 and 16). In results new to this study, the UAGG and GGGG silencer motifs were found to be important for the depolarization-induced effects on exon skipping in the primary neurons. A complete disruption of KCl-induced exon skipping was associated with point mutations in two or more of the silencing motifs (substrates E17 and T8; lanes 10–12 and 16–18). In contrast, an increase in exon skipping occurred in parallel with the depolarization treatment for the wild-type substrate (ΔEI, −17%; lanes 1–3). Point mutations in single motifs showed a similar trend in which the KCl-induced effects on exon skipping were disrupted at various levels (substrates E8, E9, and D0; lanes 4–9 and 13–15). We next tested the role of exonic enhancer motifs in the same fashion, since these motifs function generally to antagonize exon silencing. The sequence, type, and position of the exonic enhancer motifs and their inactivating mutations are shown in B. Of the six mutants, E2, E3, E4, and E5 showed reduced basal levels of CI exon inclusion in neurons, and the response of these exons to KCl-induced depolarization was either similar to (E6 and 5 m1, lanes 34–39) or reduced (E2, E3, E4, and E5, lanes 22–33) compared to the wild-type CI cassette (lanes 19–21). These results suggest that sequences outside of the UAGG motifs in the exon may also play a role in the induced splicing silencing response.
To validate these results, we asked whether a multicomponent UAGG silencing motif code is sufficient to confer sensitivity to KCl-induced depolarization; we introduced UAGG motifs into exon 5 of the gene encoding the human Dip13 beta adapter protein (DIP13B_HUMAN). This exon was chosen based on the use of algorithms that predicted strong 5′ and 3′ splice sites, a moderate number of exonic enhancer motifs (n
= 10), and a lack of known exonic silencing motifs (MaxEntScan, ACEScan, and ESEFinder). Another attractive feature was the clustered arrangement of predicted enhancer motifs that allowed for the insertion of UAGG motifs at multiple discrete positions. The region containing exon 5 and its flanking splice sites was cloned into a chimeric splicing reporter by replacing the middle exon of the previously described SIRT1a plasmid [29
], and the promoter was replaced with CaMKII_22 to direct expression in neurons (A, DIP13_E5). When expressed in neurons, the DIP13_E5 splicing reporter was nearly insensitive to the excitation treatment (ΔEI, 0%; B, lanes 1–3). In order to test the role of the silencing motif pattern in the response to excitation, we next introduced three exonic UAGG motifs and a 5′ splice site GGGG motif (DIP_3aG). Introduction of the silencing motif pattern indeed generated an exon-skipping pattern in resting neurons (61% exon inclusion), and in the presence of KCl treatment to induce excitation, exon inclusion progressively decreased to 45% (ΔEI, −16%; lanes 4–6). In another variant, DIP_E2, point mutations were introduced into the exon to destroy overlapping SC35 and ASF-SF2 motifs near the 3′ end of the exon (A). For DIP_E2, exon inclusion decreased in resting neurons, and the response to excitation was similar to that observed for DIP_3aG (B, lanes 7–9). Thus, we conclude that a multicomponent UAGG silencing motif code is sufficient to generate induced exon skipping in response to KCl-induced depolarization.
Transfer of Multicomponent UAGG Silencing Motif Pattern Confers Sensitivity of a Constitutive Exon to Depolarization in Primary Neurons
Biochemical Analysis of Protein Binding to Silencing Motifs in Nuclear Extracts from Resting and Excited Cultures
We reasoned that factors known to be involved in splicing silencing via UAGG silencing motifs in resting cells would be likely candidates to mediate the induced splicing silencing response in excited neurons. Alternatively, excitation might weaken the role of factors involved in antagonizing this silencing mechanism. We initially focused on hnRNP A1 to probe for changes in regulatory factors, because our previous work demonstrated direct binding of this factor to the UAGG motifs in HeLa nuclear extracts, and because its silencing effect in vivo was dependent upon the intact motif pattern.
To attempt to visualize whether changes in hnRNP A1 accompany KCl-induced depolarization, RNA-protein binding assays were used as sensitive readouts. In these experiments, 5-d cortical cultures were treated with KCl in the medium to induce depolarization, and cells were harvested for preparation of small-scale nuclear extracts (see Materials and Methods
). The time of KCl treatment was shortened from 24 h to 12 h based on the expectation that any excitation-induced alterations to splicing factors should precede the splicing pattern shifts themselves.
Ultraviolet (UV) crosslinking was used to monitor direct protein binding to 5′[32
P]-labeled RNA oligonucleotides under splicing conditions, and the crosslinked proteins were resolved by SDS-PAGE. A 22-mer containing three UAGG motifs showed increased crosslinking of a 34-kDa protein, suggestive of hnRNP A1, in nuclear extracts prepared from KCl-induced versus mock-treated samples (A, lanes 1 and 2). Increased crosslinking of the 34-kDa protein in the KCl-induced extracts was observed in eight different nuclear extract preparations, and the fold increase was in the range of 1.5- to 2.5-fold. To determine whether hnRNP A1 was responsible for the observed increase in crosslinking, a size-matched control RNA with point mutations in the UAGG motifs (A1 mutant) was tested in parallel reactions. Crosslinking of the 34-kDa protein to the A1 mutant oligo was abolished, demonstrating that intact UAGGs are essential for binding. To confirm the identity of the 34-kDa protein, three UV crosslinking reactions equivalent to that shown in lane 2 were combined and immunoprecipitated with monoclonal antibody 9H10, which is specific for hnRNP A1. These results showed that, relative to the control antibody, the 34-kDa crosslinked protein was partitioned into the pellet (and depleted from the supernatant) in the presence of 9H10, supporting its identification as A1 (Figure S2
). For comparison, parallel crosslinking reactions with distinct RNA oligos specific for human Tra2 and ASF/SF2 were also tested, but these proteins showed no detectable change in binding in extracts prepared from the KCl-induced versus mock-treated cultures.
Biochemical Analysis of Protein Binding to UAGG Silencer Motifs as a Function of Cell Excitation
Enhanced hnRNP A1 binding was also observed for the full-length CI cassette exon in UV crosslinking reactions prepared with nuclear extracts from KCl-induced cultures (A. lanes 9 and 10). Note that the full-length CI cassette exon tested here contains three UAGGs distributed throughout the 111-nucleotide exon, which is a less-concentrated motif arrangement relative to the A1 oligo. In contrast to hnRNP A1, higher molecular weight proteins in the 57- to 100-kDa range in these samples serve as reference comparisons, since these displayed similar crosslinking levels in the two nuclear extracts (lanes 9 and 10, Ref bands). The identity of hnRNP A1 was verified by competition experiments with unlabeled A1 oligo, which reduced crosslinking to the 34-kDa protein in a concentration-dependent manner (unpublished data).
To verify the change in hnRNP A1 binding observed in the UV crosslinking assay, affinity selection was used as a complementary method. For this experiment, the full-length CI cassette exon with three UAGGs was subcloned upstream of the M3 hairpin to provide an affinity tag that was specific for the MS2-MBP fusion protein [30
]. The CI cassette exon was subcloned without the M3 hairpin as a control. To assay for hnRNP A1, one half of the eluted samples were separated by SDS-PAGE and immunoblotted with the 9H10 antibody. Parallel blots of the second half of the samples were developed with an antibody specific for ASF/SF2. Binding of hnRNP A1 was dependent upon the presence of the M3 hairpin (B, lanes 13–16), and an increase (1.5-fold) in A1 binding to the M3_E18 substrate was observed in the KCl-induced versus mock-treated extracts (lanes 13 and 14). This difference in A1 was similar in the input samples prior to RNA binding (lanes 11 and 12). The increase in the level of A1 protein was similar to the input samples prior to RNA binding. In contrast, ASF/SF2 showed no change in binding to the M3_E18 substrate in the two samples, nor was there an observable difference in ASF/SF2 in the input samples (lanes 17–22).
To confirm the difference in nuclear levels of hnRNP A1, quantitative Western blotting was used to measure A1 levels in nuclear extracts from resting and excited cultures relative to known levels of recombinant A1 (C). These results verify that an increase in nuclear A1 protein levels in the cultures is associated with the depolarization treatment.
Evidence for the Role of the NMDA Receptor and Calcium Signaling in the Induced Splicing Response
We would expect cell excitation initiating at the cell membrane to communicate changes to splicing factors in the nucleus via calcium-mediated signal transduction pathways. In the absence of antagonists, the NMDA receptor channel opens in response to membrane depolarization and serves as the major conduit for calcium entry into neurons. Because NMDA receptors are functionally intact in neurons in these cultures, we wished to take advantage of this property of the culture system to ask whether NMDA receptors play a role in the induced effects on splicing observed here. To address this question, we expressed the CI wt0 splicing reporter in 5-d cortical cultures under the control of the CaMKII_22 promoter, and the cultures were pre-incubated with antagonists specific for the NMDA receptor before KCl-induced depolarization. MK801 and AP5 antagonists were used for these experiments because of their known specificity and effectiveness in inhibiting the NMDA receptor calcium channel in cortical cultures [31
]. AP5 binds competitively to the extracellular glutamate (NMDA) site of the NR2 subunit, whereas MK801 binds to the channel itself, blocking ion flow. The exon-skipping response of the CI cassette exon was strong in the control (mock treated) samples (A, lanes 1–3), but in the presence of either antagonist, this response was attenuated in a dose-dependent manner (lanes 4–15). Relative to the control sample (ΔEI, −25%), exon skipping decreased in the presence of MK801 (ΔEI, −10%) and in the presence of AP5 (ΔEI, −4%). In contrast, when cultures were pre-incubated with bicuculline, which is an antagonist of the GABAA
receptor, only slight effects were observed on induced exon skipping (ΔEI, −22%; lanes 16–21) relative to the control (ΔEI, −25%; lanes 1–3). Thus, to a first approximation, the effects of MK801 and AP5 implicate a role for the NMDA receptor in mediating the exon-skipping response of the CI cassette exon in this system.
Test of Inhibitors of NMDA Receptors and Signaling Pathways on the CI Cassette Exon-Skipping Response in Excited Neurons
To extend these results, we used cell-permeable inhibitors to ask whether calcium-mediated signal transduction pathways downstream of NMDA receptors are involved in the induced splicing silencing response of the CI cassette exon. Relative to the mock-treated control, the exon-skipping response was strongly inhibited in the presence of 15 μM KN93, which is an active site-based inhibitor of CaMK I, II, and IV (B, lanes 22–24 and 40–42), and this effect was lessened when the inhibitor concentration was reduced to 3 μM (lanes 37–39). The control compound, KN92, had little or no effect in these experiments (P. An and P. J. Grabowski, unpublished data). We also found that the exon-skipping response was inhibited by H89, which has been widely used as an inhibitor of PKA (lanes 31–36). In an attempt to confirm this effect, the inhibitor KT5720, which is reported to have a higher specificity for PKA, was also tested. Interestingly, KT5720 had little or no effect at 2.5 and 10 μM concentrations (lanes 25–30), indicating that the inhibitory effect of H89 on splicing silencing observed here may involve a distinct pathway (or pathways).
To determine if additional calcium channels could play a role in the splicing response of the CI cassette exon, we tested the effects of specific antagonists of the AMPA/kainate receptor and voltage-gated calcium channels (L- and N-type) (Figure S3
). Relative to control samples, the exon-skipping response was reduced, but not eliminated, in the presence of nimodipine (antagonist of L-type calcium channels) and conotoxin (N-type calcium channels), whereas CNQX (AMPA/kainate receptors) showed little or no effect. Taken together, these results are consistent with roles for multiple calcium channels in the induced splicing response of the CI cassette exon in neurons.
Depolarization Alters Splicing Patterns of Additional Transcripts with Synaptic Functions
Next, we expanded the analysis of endogenous transcripts in the cortical cultures to test whether additional alternative cassette exons were responsive to depolarization, and if so, to determine whether exonic UAGG silencing motifs were required. We were also prompted to examine transcripts that encode protein components with synaptic functions, since the NMDA NR1 receptor and other calcium channels were implicated in the splicing response of the CI cassette exon from the experiments described above. Of 14 new exons tested, seven showed a significant exon-skipping response, five showed little or no response, and two showed an increase in exon inclusion ().
Exon-Skipping Responsiveness of Fourteen Endogenous Exons in Cortical Cultures During Depolarization Treatment
Strong exon-skipping responders were found to contain hnRNP A1–type silencing motifs in or near the responsive exon (). Exon 3 of hnRNP H3 and exon 2 of the RNPS1 transcripts (ΔEI, −32% and −25%, respectively) contain either one exonic UAGG and a 5′ splice site GGGG motif (H3), or two exonic UAGGs (RNPS1). Exon 35 of PLCβ4 also showed a strong response (ΔEI, −26%), and this exon contains two TAGA motifs, which reflects a four of six match to hnRNP A1 motifs reported for SMN2 exon 7 and human CD44 v5 exon [33
]. These motifs cannot be the sole determinants of the response, however. Exon 4 of hnRNP H1, which is identical in size (139 nucleotides) and highly homologous to exon 3 of hnRNP H3, showed little or no response (ΔEI, −3%) in the same samples. Interestingly, there is one exonic UAGG and a GGGG tetramer near the 5′ splice site of this exon, but it lacks a third silencing motif (exonic GGGG) that is found in exon 3 of hnRNP H3. Moreover, constitutive exon 8 of the MEN1 transcript remains entirely unresponsive even though this exon contains two exonic UAGGs and a 5′ splice site proximal GGGG motif.
Summary of Exon-Skipping Responsiveness and UAGG Code
Of the exons that showed little or no response, exon EN of clathrin light chain B (CLCB)
was the most interesting. In a previous report, this exon showed increased exon skipping in the cortex and hippocampus of rats treated with pilocarpine [13
]. In cortical cultures under the conditions tested here, however, exon EN showed a stable splicing pattern even though it is a well-skipped exon (exon inclusion, 30%). Other alternative exons with a predominant skipping pattern that were unresponsive in these cultures include exon 21 of MAP4k4, exon 8 of hnRNP A1, exon N1 of GRIN1,
exon 7 of Chl,
and exon 7 of Agrin
. These exons, including EN of CLCB,
lack hnRNP A1 motifs. Thus, the exon-skipping responsiveness of certain exons to depolarization treatment cannot be explained solely by a weakening of the general splicing machinery that causes all skipped exons to undergo stronger exon-skipping responses.
Finally, two exons showed moderate increases in exon inclusion during the depolarization treatment. These included exon 9 of the GABAA receptor γ2 transcript, and the STREX exon of the c-slo transcript (ΔEI, 18%, and 12%, respectively). These results suggest that some of the biochemical changes resulting from the depolarization treatment in neuronal cells can, in principle, increase exon inclusion.