Ule et al., Supplementary Fig Legends Supplementary Figure S1: General scheme to define and validate the RNA map and relate it to the mechanisms of Nova-dependent splicing regulation. Prior work established YCAY clusters as legitimate target sequences for analysis, based on RNA selection, biochemistry/X-Ray crystallography1-7, and new methods for identification and genetic validation of Nova splicing targets2, 6, 8. In the current paper, these targets were first analyzed by a computational procedure to define the Nova RNA map of splicing regulation, which in turn predicted the action of Nova on new splicing targets (top triangle). The RNA map was related to the mechanism of Nova action through biochemical studies in a reconstituted splicing system in vitro (circle). Two examples illustrate Nova inhibition of exon inclusion by binding NESS2 element and blocking U1 snRNP assembly on the pre-mRNA, and Nova upregulation of exon inclusion by binding NISE2/3 element to enhance spliceosome assembly. These results were generalized by quantification of splicing intermediates in Nova dKO brain, which showed that the RNA map predicts the in vivo site of Nova action in all 19 pre-mRNAs where we detected an asymmetry in Nova’s action on removal of the two introns flanking the alternative exons. Supplementary Figure S2: Analysis of the positions of YCAY motif enrichment in previously validated Nova-target pre-mRNAs. a, An outline of four regions within the pre-mRNA that were analyzed to define the RNA map. The peaks demonstrate the enrichment of YCAY motifs in pre-mRNAs with exons down-regulated (blue) or upregulated (red) by Nova. b, YCAY motif frequency was calculated in 60 nucleotide sequences at 20 nucleotide intervals on the pre-mRNAs containing Nova-regulated exons. Mouse sequences of 34 pre-mRNAs with exons up-regulated by Nova (Nova +), 21 pre-mRNAs with exons down-regulated by Nova (Nova -) and 3865 pre-mRNAs with control exons (control, not regulated by Nova, as shown by a previous study9) were analyzed. In addition, for each 60 nucleotide window in Nova-regulated pre-mRNAs, 25 shuffled sequences were generated to generate over 500 sequences for each analyzed position in pre-mRNAs with 1 Ule et al., Supplementary Fig Legends exons up-regulated by Nova (Nova + shuffle) and pre-mRNAs with exons downregulated by Nova (Nova - shuffle). Each dot on the graph represents the average values (within each group) of 60 nucleotide sequences that are located at the same distance from the splice site. The error value of the control pre-mRNAs was calculated by generating 100 random sets of 20 pre-mRNAs, and calculating standard deviation between the average values of each set. The diagram below the graph illustrates the part of the premRNA that is being analyzed in the graph. At each significant peak of YCAY enrichment in Nova-regulated versus control pre-mRNAs, the fold enrichment of YCAY motifs and p-value (two-tailed t-test, unequal variance) is shown. Supplementary Figure S1: Analysis of the YCAY cluster characteristics. The highest score YCAY cluster was found within previously validated Nova regulated pre-mRNAs (|YCAY cluster score| > 0.5, the clusters sequences are shown in Supplementary Table S2). a, Distribution of the YCAY cluster widths (nucleotides from the beginning of the first YCAY to the end of the last YCAY motif are counted). b, For each YCAY within the cluster, the distance from the next closest YCAY motif was measured, and plotted in the distribution of inter-YCAY distance within the clusters. (-1 stands for cluster YCAYCAY, 0 for YCAYYCAY, 1 for YCAYNYCAY, etc.). Majority of YCAY motifs are 2 or less nucleotides apart, and 90% of are 6 or less nucleotides apart. c, Analysis of the YCAY cluster conservation. To analyze the conservation of YCAY clusters in Nova-target pre-mRNAs, we compared it with conservation of YCAY clusters located at corresponding positions within control pre-mRNAs (containing alternative exons not regulated by Nova, as determined by a previous study9). 37 YCAY clusters with |YCAY cluster score| > 0.5 were found within 200 nucleotides of the alternative exon or flanking exons in Nova-regulated pre-mRNAs, and 152 were found in 1388 control pre-mRNAs in mouse genome. In control pre-mRNAs, 65% of the exonic YCAY clusters and 33% of intronic clusters are conserved. Higher conservation of exonic clusters would be expected due to coding constraint. 97% of the YCAY clusters in Nova- 2 Ule et al., Supplementary Fig Legends regulated pre-mRNAs are conserved, and there is little difference between the conservation of intronic and exonic clusters, suggesting that the main conservation constraint to these sequences is their role in binding Nova. Independent analysis of these exons and their corresponding mouse and human cDNAs suggests that those exons that contain conserved YCAY clusters are similarly regulated in human; therefore, we estimate that ~97% of the exons found to be regulated by Nova in mouse are conserved alternative exons, regulated by Nova also in human pre-mRNAs. Supplementary Figure S4: Analysis of the predictive accuracy of net YCAY cluster score at different stages of the algorithm. The graphs in a-f show analysis of 48 Nova-regulated exons that were identified by previous studies2, 6, 8, in addition to 1388 control exons. 1388 control exons were a subset of the 4738 exons that showed little or no splicing change by splicing microarray analysis of Nova2–/– brain9, where we were able to collect orthologous sequences from all the regions on the RNA map from zebrafish, Xenopus, chicken, opossum, dog, mouse and human genomes. I9 (splicing change), as determined using splicing microarray or RTPCR2, 6, 8, is shown on the x axis, and the net YCAY cluster score, calculated at different stages of the algorithm, on the y axis. A positive net YCAY cluster score predicts Novadependent exon inclusion, and negative score predicts Nova-dependent exon skipping. To analyze the accuracy of the net YCAY cluster score at each stage, we have analyzed two values. First, as an estimate for accuracy (regular text in the figure), we determine the cutoff level where roughly a 25% of the Nova-regulated exons are predicted, and calculate the percentage of exons predicted above this cutoff that are regulated by Nova (relative to control exons). Second, as a measure for efficiency (italic text in the figure), we determine a cutoff level that distinguishes between exons that are up- or downregulated by Nova (that means, a positive cutoff above which no down-regulated exon is predicted, and negative cutoff below which no up-regulated exon is predicted), and calculate the percentage of Nova-regulated exons that are predicted above this cutoff (relative to all Nova-regulated exons). 3 Ule et al., Supplementary Fig Legends a, YCAY score was calculated using only the mouse sequence, by subtracting the maximum value of a cluster in silencer regions from maximum value of a cluster in enhancer regions. Accuracy is 17% (12/72), and efficiency is 35% (17/48). b, YCAY score was calculated using only the mouse sequence, by subtracting the maximum value of a cluster in silencer regions from maximum value of either a single cluster in enhancer regions, or 2/3 of a sum of maximum clusters in NISE2 and NISE3 regions. Accuracy is 19% (14/74), and efficiency is 42% (20/48). This step exploits a common feature of Nova targets to contain YCAY clusters both in NISE2 and NISE3 region, and thus increases the number of predicted up-regulated exons, with no effect on control exons, which rarely contain two YCAY clusters in flanking intron (see Supplementary Fig. S4b) c, YCAY score was calculated using the 2x minimum value of YCAY clusters in orthologous mouse, human and dog sequences. Accuracy is 79% (11/14), and efficiency is 54% (26/48). This step shows the important of orthologous sequences for the accuracy of the algorithm, since it increases the ability to distinguish between Nova-regulated and control exons. It also increases efficiency, since it has a better ability to distinguish between exons that are up- or down-regulated by Nova. d, YCAY score was calculated using the average value of YCAY clusters in orthologous mouse, human and dog sequences + average value of YCAY clusters in orthologous opossum, chicken, Xenopus and Zebrafish sequence. Accuracy is 69% (11/16), and efficiency is 63% (30/48). Thus, this average value has a slightly lower accuracy than the minimum value (in c), but slightly higher efficiency, since minimum value doesn’t predict exons with clusters in all but one species. e, YCAY score was calculated using the 2x minimum value of YCAY clusters in orthologous mouse, human and dog sequences + the average value of YCAY clusters in orthologous mouse, human and dog sequences + average value of YCAY clusters in orthologous opossum, chicken, Xenopus and Zebrafish sequence. Accuracy is 76% (13/17), and efficiency is 63% (30/48). Thus, combining the minimum and average values gains optimum accuracy and efficiency. f, The score was calculated as in e, but only using the cluster values greater than 0.6 (values <0.6 were changed to 0). Accuracy is 76% (13/17), and efficiency is 66% (31/48). 4 Ule et al., Supplementary Fig Legends This last step removes the background noise, with no negative effect on ability to predict Nova target exons. Supplementary Figure S5: YCAY cluster position in individual Nova-regulated premRNAs. Conserved YCAY cluster score was determined for each 45 nucleotide sequence along each individual Nova-regulated pre-mRNA. The pre-mRNAs with similar pattern of YCAY cluster positions were grouped together and plotted onto the same graph. a, 9 pre-mRNAs containing YCAY clusters both in NISE2 and NISE3. b, 7 pre-mRNA containing YCAY clusters in NISE2, but not NISE3. c, 2 pre-mRNAs containing YCAY clusters in NISE3, but not NISE2. d, 2 pre-mRNAs containing YCAY clusters both in NISS1 and NESS. e, 3 pre-mRNAs containing YCAY clusters in NISS2 (and NESS) f, 4 pre-mRNAs containing YCAY clusters in NESS. Pre-mRNAs containing two YCAY clusters (shown in a and d) were 58 fold more common in Nova-regulated pre-mRNAs (35%, 9/31) than in control pre-mRNAs with YCAY clusters (0.6%, 1/158), suggesting that cooperative action via different binding sites might contribute to Nova-dependent splicing regulation. However, two binding sites do not seem to be a prerequisite, as evident by cases shown in b, c, e, f (see also Supplementary Figure S8). Supplementary Figure S6: Functions of YCAY clusters in the context of consecutive alternative exons, alternative splice sites or terminal exons. YCAY clusters predicted 5 cases of Nova-dependent regulation of consecutive alternative exons (a-c), and 4 cases of Nova-dependent regulation of alternative splice sites (d, Fig. 2d-e). a, In two pre-mRNAs, a YCAY cluster is located between two alternatively skipped exons. In both cases, the cluster acts as a NISE2 element to promote inclusion of the upstream exon (E14 in CP110 and E9 in Bcas1). Notably, E14 of CP110 is by bioinformatic criteria considered constitutive, since no cDNA or EST in the database skips E14 and in wild-type brain E14 is rarely skipped. Only analysis of Nova knockout 5 Ule et al., Supplementary Fig Legends brain shows that inclusion of this exon requires a splicing factor, suggesting that analysis of mutants in splicing factors could reveal new regulated alternative exons that do not appear alternative by bioinformatic criteria. In both CP110 and Bcas1, the YCAY cluster also acts as NISS1 to promote an isoform that skips the downstream exon (it promotes isoform E14-E15 in CP110, skipping E14a, and isoform E9-E11 in Bcas1, skipping E10). However, there is a crucial difference between the two RNAs. In case of CP110, splicing of E14 to E14a is very inefficient (see Supplementary Fig. 14g), therefore the main action of the cluster is to silence E14a inclusion (as NISS1) by promoting splicing of E14 to E15. On the other hand, Nova can promote splicing of E9 to E10 in Bcas1 (see Supplementary Fig. 14a), therefore the main action of the cluster is to enhance E10 inclusion. b, The cases of Sh2bpsm1 and Map4 demonstrate the specificity of Nova’s action at YCAY clusters, because only the exon in the vicinity of NISS2 element is regulated, but not the alternative splice sites of the flanking exons. c, The case of Dab1 exons 7b and 7c shows that Nova can silence two consecutive alternative exons when both contain a NISS2 element. d, The ability of RNA map to predict Nova’s action on alternative splice sites (NISE2 enhances and NESS1 silences a splice site) suggests local action of Nova via YCAY clusters. e, Two examples of Nova-dependent regulation of terminal exons predicted by the RNA map. All bands were confirmed by sequencing. Supplementary Figure S7: Position of NISS2, NISE1 and NISE3 elements relative to putative branch sites. The sequences upstream of the 3’ splice site containing NISE1, NISS2 or NISE3 elements in different Nova-target pre-mRNAs were aligned between mouse, human and opossum pre-mRNAs. YCAY motifs were colored in purple, and conserved YCAY clusters were labeled by purple rectangles. a, NISS2 elements are located within 5 nucleotides, or downstream of the conserved branch site (labeled by blue rectangle). 6 Ule et al., Supplementary Fig Legends Two examples of NISS2 elements are the YCAY clusters present upstream of the alternative exons 7b and 7c (termed 555* in a previous study10) of Dab1 RNA. In human, mouse and opossum, these two exons are ~50 bp long, separated by an intron of ~90 bp. The peptide sequences encoded by the two exons are highly related, suggesting a duplication event during evolution10. This observation agrees with finding that chick and lizard mRNAs contain only the second of the two exons10. Interestingly, these species also lack one of the NISS2 clusters, suggesting that YCAY clusters have co-evolved with the exons 7b and 7c. Exons 7b and 7c are included in mRNA of proliferating neuronal precursors, but are absent from mRNA of differentiated neurons, with silencing starting in embryonal day 12 brain10. Chick retinal cultures transfected with a neuronal Dab1 isoform (lacking exon 7b and 7c, and containing another neuron-specific exon) were able to form elongated processes and activate Src family kinases, whereas transfection of the isoform containing exons 7b and 7c had no effect11. This suggests that exons 7b and 7c might encode an inhibitory region. We observe that sequence of YCAY clusters (present upstream of exon 7b and 7c) is more conserved than sequence of exons 7b and 7c, agreeing with the possibility that these exons inhibit Dab function, therefore the main evolutionary pressure acts to preserve ability of Nova to silence their inclusion in brain. b, NISE1 elements are located 13 or more nucleotides upstream of the conserved branch site (labeled by red rectangle). Note that in case of Ank3, the YCAY cluster is located at a position where it can either act as NISS1 to silence proximal branch site, or as NISE1 to enhance the distal branch site. c, NISE3 elements are generally located upstream of the conserved branch site (labeled by black rectangle). In the three cases shown, there is another conserved branch site located 5-8 nucleotides downstream of the YCAY cluster; the significance of this second branch site is unknown. Supplementary Figure S8: Three examples of single YCAY cluster within short introns acting as NISE2 or NISE3. Even though two YCAY clusters in NISE2 and NISE3 are a common feature of Novadependent exon inclusion, two clusters do not seem to be a prerequisite. Three examples are shown here of the whole intronic sequence with a single YCAY cluster. Two introns 7 Ule et al., Supplementary Fig Legends are very short, containing a single YCAY cluster roughly in the middle of the intron (Falz and Plcl3). The third example (Lrp1b) has a clear NISE2 element, but no YCAY cluster close to the distal part of the intron (no NISE3). Supplementary Figure S9: Correlation of YCAY cluster location with exon size, or with the location of a putative exonic splicing enhancer. a, 56% of exons regulated by Nova via NISE2 elements are shorter than 50 nucleotides, whereas only 14% of the rest of the Nova-regulated exon is as short. The distribution of Nova-regulated exon sizes that are flanked by NISE2 elements is significantly different from the rest of the Nova regulated exons (p=0.003, two-tailed t-test, unequal variance). b, The most common pentamer in the 9 NESS elements with highest YCAY cluster score is CACCA. The sequences of the 9 NESS elements are shown, with the CACCA motif marked in red, and the flanking YCAY motifs in blue. Supplementary Figure S10: Bioinformatic analysis of Nova-regulated exons. a, Analysis of splice site scores, as determined by MaxEnt (http://genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq_acc.html) score12 shows no significant difference in splice site scores of Nova-regulated exons as compared to constitutive exons in the same RNAs. b, Analysis of the number of enhancer and silencer elements/100 nucleotides of exonic sequence, as determined by RESCUE-ESE and FAS-ESS (http://hollywood.mit.edu/Dexon.php)13 shows a 22% decrease in the average density of enhancer elements in Nova-regulated exons as compared to constitutive exons in the same RNAs. However, neither the difference in the number of predicted enhancer nor silencer elements is significant. c, Analysis of the conserved alternative exon score as predicted by ACEScan (http://hollywood.mit.edu/Dexon.php)13 shows a significant difference in the score of Nova-regulated exons as compared to constitutive exons in the same RNAs. The ability of ACEScan to predict that Nova-regulated exons are conserved alternative exons might be due to long regions of conserved intronic sequence flanking Nova-regulated exons, which is a characteristic of conserved alternative exons12. 8 Ule et al., Supplementary Fig Legends Supplementary Figure S11: Analysis of 10 alternative exons with net YCAY score near zero. An independent analysis of 10 alternative exons with net YCAY score near zero revealed that none were regulated by Nova (comparing Nova2–/–/Nova1–/– vs. wildtype brain). Supplementary Figure S12: Recombinant Nova represses the inclusion of exon 4 in vitro. a, Schematic of the substrate used (E-I-EYCAY-I-E): exons 1 and 3 are derived from human -globin; the lines and block in gray represent introns 3 and 4 and exon 4 of Nova1, respectively. Each vertical red line represents a YCAY motif. b, Purity of recombinant Nova1 employed in the present study: tagged Nova1 protein was purified by two affinity steps (Nickel column and anti-T7 monoclonal antibody resin). An aliquot of purified protein was separated by electrophoresis on a SDS polyacrylamide gel (5-20%) and visualized by Coomassie blue staining. c, In vitro splicing assay: Radiolabeled E-I-EYCAY-I-E (lanes 2-5) and E-I-EYAAY-I-E (lanes 6-8) RNA substrates were incubated in the absence (lane 2) and presence of ATP (lanes 3 to 8) in the presence of HeLa nuclear extract in splicing conditions. Recombinant Nova1 was included in lanes 4 and 7 (0.06 µM) and 5 and 8 (0.12 µM). After 4 hours of incubation, RNA was extracted and separated by electrophoresis on a denaturing polyacrylamide gel. The gel was dried and RNA visualized by Phosphoimager. d, Reproducibility of Nova action on EYCAY-I-E RNA in vitro. This experiment is a separate reproduction (from a total of three similar experiments) of Figure 3a. EYCAY-I-E (lanes 1 to 5; vertical red lines in cartoon represent YCAY motifs) or EYAAY-I-E (lanes 6 to 10) RNA was incubated in nuclear extract in absence of ATP (lanes 1 and 6) or in splicing conditions (lanes 2 to 5 and 7 to 10). Recombinant Nova1 was included in concentrations ranging from 0.3 to 0.12 µM (lanes 3 to 5 and 8 to 10). After incubation, RNA was extracted and analyzed on denaturing polyacrylamide gel, which was then dried and exposed by Phosphoimager. e, UV-cross-linking assay. Radiolabeled RNA substrate I-EYCAY-I (1 nM) was incubated in HeLa nuclear extract without or with Nova (0.12 µM), treated with UV254 followed by RNAse digestion, separation by SDS-PAGE and autoradiography. 9 Ule et al., Supplementary Fig Legends f, Quantitation of co-transfection experiments in 293 cells. Results refer to four independent experiments; error bars represent standard deviations. While overexpression of hnRNPK had only a marginal effect on E4 inclusion (11% decrease in exon inclusion, p <1.4 X 10-5), expression of Nova1, even in the presence of an excess of hnRNPK, led to E4 exon skipping (80% decrease in exon inclusion, p<1.7 X 10-6). g, Quantitation of the psoralen-cross-linking assay. Incubation with 0.12 µM Nova results in a 2.5 folds decreased formation of the U1 snRNA cross-linking product with the labeled substrate. The mutation of YCAY to YAAY decreases the effect of Nova (1.2 folds reduction). h, Effects of Nova1 on the formation of the spliceosome. In the presence of Nova1 the formation of complexes B and C is inhibited (lanes 5 to 8). i, Analysis of spliceosomal complexes formation on EYCAY-I-E RNA by electrophoresis on an agarose/polyacrylamide native gel. In the presence of Nova, while complexes B and C were inhibited, a complex migrating at roughly the same rate as complex A was still observed. This Nova-resistant complex still formed, although with reduced abundance, when the first 15 nucleotides of U2 snRNA were digested with RNAse H (lane 3-4). No discrete complex was observed when nucleotides 28-42 (lanes 5 and 6) or both 1-15 and 28-42 (lanes 7-8) were digested. Analogously, in extracts that had been depleted of U2 snRNP, no such complex was observed (lanes 13-14). In contrast, in extracts that had been depleted of U1 snRNP (lanes 15 and 16) or where nucleotides 1-15 of U1 snRNA were digested (lanes 9-10) a complex migrating with a similar rate as complex A is still observed, even in the presence of Nova. Summarizing, we observe a Nova-resistant complex migrating at about the same rate as complex A. U2 snRNP (lanes 13 and 14) and nucleotides 28-42 of U2 snRNA (lanes 5 to 8), which span the residues required for the recognition of the branch site, are required for the formation of this complex. The first 15 nucleotides of U2 snRNA (lanes 3 and 4), which are required for the interaction of U2 snRNP with U6, are not required (lanes 3 and 4). These findings, together with the observation that this complex still forms in the absence of U1 (lanes 15 and 16) are consistent with the previous descriptions of complexes of similar mobility forming on splicing substrates where the 5' splice site was mutated or absent (Query et al., 1997)14. These data are consistent with the localized action of Nova in regulating the 10 Ule et al., Supplementary Fig Legends recognition of the 5' splice site by U1, without interference with the recognition of the branch site by U2. j, The level of digestion of U1 and U2 snRNAs was verified on an ethidium bromide agarose gel and by Northern Blot of the same gel probed with a mix of labeled oligonucleotides complementary to U1, U2, U4, U5 and U6 snRNA. Supplementary Figure S13: Analysis of Nova enhancement of exon inclusion in in vitro splicing reactions. a, The in vitro splicing of the substrate E-I-E-IYCAY-E was repeated multiple times using different preparations of HeLa nuclear extract and recombinant Nova1. Three repeats of one such experiment are shown, where the substrate (1nM) was incubated in the absence or in the presence of 60 nM recombinant Nova1. The upper panel is a longer exposure of the upper part of the gel, showing the two-intron lariats. b, the ratio of the long (L) and short (S) products of splicing was measured by Phosphoimaging from the gel shown in panel (a). c, The identity of the bands in (a) was confirmed by a de-branching assay: RNA was eluted from the gel, incubated in the presence of buffer D (see methods) or S100 cytoplasmic extract, 8mM EDTA at 30C for 45 minutes, and analyzed by denaturing PAGE. The de-branched products ("S100" lanes) correspond to the predicted size of the two introns and the alternative exon. d, RT-PCR analysis of the in vitro splicing reaction using EE-IYCAY-E as substrate (as shown in Fig 4b). RNA from the reaction shown in figure 5B was reverse transcribed and amplified using primers complementary to the first and last exons. In presence of 0.12 µM Nova1, the spliced product normalized by the unspliced precursor increases by 2.1 fold. e, Psoralen cross-linking on E-IYCAY-E. Body-labeled E-IYCAY-E (1nM) was incubated in splicing conditions in the absence (lanes 1to 6) or in the presence (lanes 7 to 12) of ATP/CP. After 25 minutes incubation at 30C, samples were treated with UV336, RNA was extracted with Trizol, and analyzed on a denaturing gel. When nuclear extract was omitted (lanes 1 and 7), close to 100% of the RNA migrates more slowly (indicated by the star sign) than the non-cross-linked RNA (compare 1 to 2, where psoralen was 11 Ule et al., Supplementary Fig Legends omitted, and 7 to 8), suggesting that most of the substrate folds in a secondary structure stabilized by psoralen (lanes 1 and 7 are shown from a shorter exposure for clarity, because the signal in these lanes is much stronger than in the lanes where nuclear extract was included, due to nuclease activities present in the extract). Specific bands appear in the presence of nuclear extract (lane 3, indicated by arrows). These bands are dependent on the annealing of snRNA U1 and U2 to the RNA, as the DNA oligonucleotide-RNase H-mediated digestion of the first 15 residues of U1, or of the nucleotides 28 to 42 of U2 abolishes them (compare lane 3 to 5 and 6; bands are indicated by arrows). None of these bands are affected by the presence of Nova (lanes 4 and 10). The presence of Nova drastically reduces the nuclear extract-independent bands (indicated by a star) both in the absence and in the presence of ATP. Compare bands 4 to 3 and 10 to 9). These results suggest that Nova affects the splicing of this substrate not by directly modulating the annealing of snRNP U1 and U2, but by affecting the tri-dimensional folding of the substrate RNA. Examples of secondary structures of RNA substrates affecting the outcome of splicing have been reported in the literature (reviewed by Buratti and Baralle, 2004)15. Supplementary Figure S14: Analysis of splicing intermediates of Nova-regulated RNAs. Quantification of splicing intermediates, which contain one of the introns flanking the alternative exon, and two spliced exons. RNA from Nova dKO relative to wild-type littermate brain was used to quantify the effect of Nova on removal of the two introns flanking each alternative exon. The radioactively labeled, RT-PCR amplified RNAs were separated on denaturing polyacrylamide gels. The abundance of splicing intermediate can be compared to the pre-mRNA. In addition to the primer pair that amplifies the splicing intermediate, each RT-PCR contained two additional primers to monitor the efficiency of all primers (except of the cases where the intron is shorter than 1.5kb, where the same primer pair amplifies both the intermediate as well as the pre-mRNA: see panel a, Ddr1). We also used real-time PCR to quantify Nova’s effect on the accumulation of splicing intermediates, which gave similar data, and was also able to quantify effects onto intermediates that were undetectable on the gel. Data from real-time PCR is shown as 12 Ule et al., Supplementary Fig Legends PCR Baseline subtracted CF RFU (Y axis) versus PCR amplification cycle (X axis). The graphs show quantification of bands representing splicing intermediates, normalized to the pre-mRNA bands. Since the two splicing intermediates represent the two alternative pathways leading to exon inclusion, we have normalized the added value of both intermediates to 1. When Nova promotes exon inclusion, the added value of bands in WT samples was normalized to 1, and when Nova silences exon inclusion, the added value in dKO samples was normalized to 1. Color-coded objects present between each gel piece represent the pre-mRNA and splicing intermediates. The small triangles indicate the position of the primers used for RT-PCR. The circles represent Nova bound to YCAY clusters. Each RT-PCR experiment was done in biologic triplicate and the identity of bands was verified by sequencing. a, In the Ddr1 pre-mRNA containing NISE1 element, Nova promotes splicing of the 5’ intron, increasing the amount of 5’ splicing intermediate. No significant effect of Nova is seen on Smarcc2 pre-mRNA (possibly due to a smaller effect of Nova on the fold change in final isoform ratio, see Table 1). b, Analysis of 6 pre-mRNAs containing NISE2 elements. c, Analysis of 10 pre-mRNAs containing both NISE2 and NISE3 elements or just NISE3 (Gabrg2). In b-c, Nova promotes primarily splicing of the 3’ intron, increasing the amount of 3’ splicing intermediate. d, Analysis of 2 pre-mRNAs containing NISS2 elements, where Nova primarily decreases the amount of 5’ splicing intermediate. e, Analysis of 2 pre-mRNAs containing NESS1 element, where Nova primarily decreases the amount of 5’ splicing intermediate. f, Analysis of 3 pre-mRNAs containing NESS2 element, where Nova primarily decreases the amount of 3’ splicing intermediate. g, Analysis of 2 pre-mRNAs containing two elements on both sides of the exon (NISS2 and NESS2 element in case of Gphn E7a, and NISS2 and NISS2 in case of Dab1 exons 7b and 7c) shows symmetric action of Nova on both the 5’ and 3’ splicing intermediates. h, Analysis of 3 pre-mRNAs containing NISS1 element showing Nova–dependent decrease in the amount of 3’ splicing intermediate. Assuming that Nova acts locally, the NISS1 cluster should act to locally enhance recognition of the 5’ splice site of the exon 13 Ule et al., Supplementary Fig Legends upstream of the alternative exon. Such action is evident in case of CP110, where the NISS1 YCAY cluster (which silences inclusion of exon 14a) also acts locally as NISE2 element to promote inclusion of exon 14 (Supplementary Fig. S6a). Analysis of splicing intermediates shows that joining of CP110 exon 14 to 14a is not efficient, therefore the NISS1/NISE2 element promotes joining of exon 14 to exon 15. Thus, it is possible that the decrease in the 3’ splicing intermediate of exon 14a is an indirect result of the enhancing action of Nova that promotes joining of exon 14 to 15, and thereby skipping of exon 14a. However, the effect on the 5’ splicing intermediate varies, suggesting that the mechanism of NISS1 action may depend on other elements in the pre-mRNA (for example, Rap1 pre-mRNA contains a high score, conserved YCAY cluster in the middle of the 5’ intron, which may act together with NISS1 to inhibit splicing of the 5’ intron; data not shown). i, Quantification of the Nova-dependent fold change in the amount of 5’ and 3’ splicing intermediates for all of the pre-mRNAs shown in panels a-h References for Supplementary figure legends: 1. 2. 3. 4. 5. 6. 7. 8. Buckanovich, R. J. & Darnell, R. B. The neuronal RNA binding protein Nova-1 recognizes specific RNA targets in vitro and in vivo. Mol Cell Biol 17, 3194-201 (1997). Jensen, K. B. et al. Nova-1 regulates neuron-specific alternative splicing and is essential for neuronal viability. Neuron 25, 359-71 (2000). Jensen, K. B., Musunuru, K., Lewis, H. A., Burley, S. K. & Darnell, R. B. The tetranucleotide UCAY directs the specific recognition of RNA by the Nova Khomology 3 domain. Proc Natl Acad Sci U S A 97, 5740-5 (2000). Lewis, H. A. et al. Sequence-specific RNA binding by a Nova KH domain: implications for paraneoplastic disease and the fragile X syndrome. Cell 100, 32332 (2000). Dredge, B. K. & Darnell, R. B. Nova regulates GABA(A) receptor gamma2 alternative splicing via a distal downstream UCAU-rich intronic splicing enhancer. Mol Cell Biol 23, 4687-700 (2003). Ule, J. et al. CLIP identifies Nova-regulated RNA networks in the brain. Science 302, 1212-5 (2003). Dredge, B. K., Stefani, G., Engelhard, C. C. & Darnell, R. B. Nova autoregulation reveals dual functions in neuronal splicing. EMBO J 24, 1608-20 (2005). Ule, J. et al. Nova regulates brain-specific splicing to shape the synapse. Nat Genet 37, 844-52 (2005). 14 Ule et al., Supplementary Fig Legends 9. 10. 11. 12. 13. 14. 15. Ule, J. et al. Nova regulates brain-specific splicing to shape the synapse. Nat Genet 37, 844-52 (2005). Bar, I., Tissir, F., Lambert de Rouvroit, C., De Backer, O. & Goffinet, A. M. The gene encoding disabled-1 (DAB1), the intracellular adaptor of the Reelin pathway, reveals unusual complexity in human and mouse. J Biol Chem 278, 5802-12 (2003). Katyal, S. & Godbout, R. Alternative splicing modulates Disabled-1 (Dab1) function in the developing chick retina. Embo J 23, 1878-88 (2004). Yeo, G. W., Van Nostrand, E., Holste, D., Poggio, T. & Burge, C. B. Identification and analysis of alternative splicing events conserved in human and mouse. Proc Natl Acad Sci U S A 102, 2850-5 (2005). Holste, D., Huo, G., Tung, V. & Burge, C. B. HOLLYWOOD: a comparative relational database of alternative splicing. Nucleic Acids Res 34, D56-62 (2006). Query, C. C., McCaw, P. S. & Sharp, P. A. A minimal spliceosomal complex A recognizes the branch site and polypyrimidine tract. Mol Cell Biol 17, 2944-53 (1997). Buratti, E. & Baralle, F. E. Influence of RNA secondary structure on the premRNA splicing process. Mol Cell Biol 24, 10505-14 (2004). 15